I recently started integrating tmux and vim into my workflows. Obviously, it was a huge efficiency booster. What are some other tools (custom built or off-the-shelf), hotkeys, workflows that were game-changing for you as a dev? I'm interested in hearing about anything and everything (stuff from ctrl+l to clear the terminal to little-known git commands to larger-scale strategies like CI/CD)
Obs studio. I record myself while coding anything. I pretend to be a bigshot coding streaming sensation (even if its just for me). Its fun as well as very helpful in so many ways.
1. It helps me stay focus on the task at hand. One recording for each task
2. It lets me practice how to articulate stuffs. Its like blogging but ephemeral (because i wont upload this)
3. It helps me get motivated. Cant let my "thousands" of viewers down
4. It lets me review my blunders so i know how to watch out for them in the future
5. Its fun. I can let off some steam because i can curse all i want, my audience is 18+ and fine with some "sentence enhancers". I am Filipino but i record in English so i get that extra bonus practice.
Alternatively, if you don't care about having an audience, you can just do private livestreams directly to YouTube via OBS. After your stream is concluded, it saves the livestream as a VOD. With this method, you don't have to worry about storing large video files when recording yourself.
Also, because it's YouTube, you inherit all of the benefits YouTube provides. You get the added benefit of having closed captioning for your videos after processing. You can watch through your recordings at 2x speed. And of course, you can share VODs to other people if you ever need to.
I've been treating YouTube as a personalized streaming website with free, unlimited storage.
If you do end up doing this, make sure to avoid any copyrighted content (Netflix, official music, etc.), otherwise YouTube will copyright strike your channel.
I also considered this in the past but sometimes, I slip into procrastination mode unintentionally and might check my bank account (or pornhub) once in a while. So thinking my video is uploaded to somebody else's computer gives me the shivers.
It seems the point of doing this is it stops you procrastinating, because you're pretending to be a hot shot twitch streamer. If it were a live stream, you wouldn't start watching porn, you'd stay focused.
Basically the productivity tip is to pretend you're already productive.
Fair point. But sometimes the pretending part fades away when I get to the "flow". When that happens, I usually forget OBS was even running way after the task was done. If it were Twitch stream, I imagine somebody from the chat window would remind me to put my d* away
Thank you for posting this. A couple of months ago I built out the design then all the FE code for a prototype app, and recorded the whole process as J was doing it, mainly as an experiment to see where o could improve a few aspects of my designing/developing.
It was all a huge faff as I was just recording short screen captures using QuickTime. But it's definitely the most productive I've been for a very long time, as well as ending up with designs better than anything else I've done for the last year. I did it at work in the office over a few days, so couldn't really talk my way through it (which I think would have improved things still further). I haven't repeated the technique because of the (non programming) effort involved, but Obs seems very good, and something that should push me to try to implement the technique a lot more.
One thing that helped me was screen recording my editor session and then re-watching it on the highest playback speed. You start to get a feel for the arc of laying out the code, what refactoring operations look like. Most code is selection, map, map, filter, reduce, re-join, etc. Watching your actions get played back, allows you to re-think those thoughts but in the third person. Mirrors help, video helps, we are not ourselves, once you realize that, optimization becomes much easier.
> Mirrors help, video helps, we are not ourselves, once you realize that, optimization becomes much easier.
It's funny, in two of the areas I've worked in (tailoring and illustration) I was taught techniques that are analogues to this. With illustration, if you draw from a reference, one of the ways to improve accuracy is to turn the reference image upside down, which basically stops your brain seeing it as the thing it is an image of. Instead it becomes a structure of shapes, it's easier to draw. With tailoring, when cutting on a model/dummy, I was taught to continually check the work in a mirror instead of always looking directly at it -- again, I think a way to stop the brain automatically filling in structures and shapes, to force it to see the piece as a whole from a different perspective.
I was trained (and worked, for a short time) in tailoring/fashion, and was also trained and worked (for a longer time) as a designer and illustrator. I was just taught to do this as a matter of course (by tutors and by technicians). It's just finding ways to trick your brain into actually thinking about things instead of relying on (habitually ingrained) automatic movements which tend to cause you to miss obvious errors.
Re illustration, this is quite common when explaining how to draw from a reference picture. Not at any detailed length as far as I can ever remember, literally just "turn the photograph upside down when you're drawing from it" (doesn't really need much explanation, that just how it works). As far as introductory books go I can remember "Drawing on the Right Side of the Brain" goes through this.
Re tailoring, the issue is that you tend to fuck up the balance of things in 3D when you're sitting staring at specific bits of it, even when you step back you only tend to see what you want if you've been at it for a while. You need to be able to see it from a. different angles at the same time, and b. not just directly through your eyes
The non-dominant hand will work for what you're doing, it's a good thing to do I think (in a way I wish I'd been taught to practise using my right more when I was a child). The reference image technique is specifically to increase accuracy though, so slightly different aim
I take it they're speaking generally, as that covers all languages: map something to data, take that data and do something to it, map it back out to something a human or computer wants.
I loved this idea. Can you share your Obs defaults/workflow for this? By example, I work 8hs per day, how much space all this audio/video recording will take? What video/resolution/etc? Thank you!
Honestly for me it was just the act of being able to see my picture and my screen moving inside OBS, not even the fact that it was recording. I tried streaming myself live coding on Twitch a couple times and it was a blast. I felt like I was thinking so much more clearly. I didn't save the recordings—just the act of it all was worthwhile.
the idea is to make a video that you can watch yourself. 8 hours of video isn't very enticing to watch. i suggest organizing the recordings by feature or task.
the only other thing I considered setting up in OBS is to make it work with Spotify to get a great bg music.
That's a great idea. I have done this with vlogs/podcasting and even blogging (created several fake blogs over the years, even in a journal on paper), but never while coding or otherwise doing stuff on the screen. Good stuff.
So basically, when procrastinating, we tend to justify our inactions: just 1 more minute on Twitter, just one last scroll on my newfeed, lemme just check this subreddit really quick.
With reverse procrastination, you trick your brain into doing your intended task: Ok lemme just open Visual Studio Code and edit this one file then im good, I'll just change this one line of code, I'll just do 5 pushups then i'll stop. Then when you're on your 5th pushup, you say "surprise mr lazy pants, im actually doing 10!"
Once you get to do that one task, try to trick again your brain into doing another simple, easy task and before you know it youre in "the zone".
Maybe it depends on the person, but for me personally, it works pretty good.
Perhaps OP is asking for tools, but no amount of dev tools can help if you dont have the motivation to begin.
One thing you can try: actually stop after those 5 push-ups. Actually stop after editing that line of the code.
The idea is that I need to build tolerance for that task. If you keep at it, those 5 push-ups will become so easy that you will feel compelled to do more.
If it's not working, then there's another reason behind the resistance. Maybe what you're trying to do feels pointless. Try to notice your feelings and internal dialogue when you're trying to do that particular thing.
Or just make the task smaller and smaller. Can’t finish 5? Do 1! Too much? Just get on the floor and hold the position. This strategy can’t fail, as you can always reduce your goal (and next time have to reduce it less)
I agree and it's important, at the same time, not to trick yourself or push yourself. If you don't feel like doing more, don't. With time, you'll want to do more.
I usually have my current project opened in Visual Studio Code at a separate desktop, so I can just swipe right and start coding. If there is an easy task I can complete quickly(not a lot of thinking) to hook me in it is pretty efficient.
who says Twitter/HN doesnt constitute a context switch? In my experience, though very boring, sitting and waiting for the tests to pass is much more productive than even the slightest diversions. I believe this is due to focus and memory.
Interesting tip. Additionally, I believe attacking the cause of procrastination can be helpful in the long term. Just asking "Why do I need to check my Twitter feed?" and "Is it more important than the task I am currently tending to?" can go a long way in reducing procrastination.
I read this book on micro tasks by the guy that started the one pushup challenge. I have been doing pushups every day for over a year now. It takes a while to make it a habit, but it does work.
The things I've benefited from most are not tools but practices.
- Question the work
Always be questioning whether the thing you're doing really needs to be done. Is there a way to not do it? Is there a way to do something better instead?
- Park downhill
Before putting a project or incomplete task away, make notes of what the next thing was that you were going to work on. This lets you bypass that 10 minute orientation getting back into the project the next time you pick it up.
- Keep a single todo list outside of your head
You can only have one master, so keep a single todo list in order of priority. Whenever something pops up on your radar, put it on the list in the right spot and immediately forget about it. Check the todo list a few times a day to make sure you're working on the right thing, and whether you need to communicate that something low on the list won't get done in time. Aggressively try to weed out the list and remove as many items as possible, preferably by not having to do them (by reassigning, by replacing or by reprioritizing).
Benefits: (1) not forgetting to do any work, (2) free up your mind to focus on work instead of meta-work, (3) clear priorities, (4) detect early when work is going to get delayed.
To combine this with park downhill I either keep the per-task notes on what to do next right in the list (in a details field of the todo item), or I keep it somewhere else and have just a single top-level item "finish X" in the master list.
Semi-related to “park downhill” is something I do at the end of the day if I’m still working on a piece of code - I’ll write a simple next task in a non-comment so it breaks the syntax. Then the next day I can dive in and “fix” the code which, at least for me, helps get in the groove.
Sometimes it's already a compromise between too many people who questioned it, and nobody wants to discuss it any more. Sometimes the questioning is more work than actually doing it. Sometimes you get a reputation of only being in the way.
Make sure the work has been questioned thoroughly, but that doesn't necessarily mean by you.
I love that last line. I think sometimes developers can feel like they can be the hero, stopping the business from going down the wrong path simply because nobody else asked the question. They may even be right! But question your motives. Are you asking questions and throwing up roadblocks to make yourself look and feel important? Or is it purely to make the best business decision?
That's not a leading question. Honestly, both of those goals can be important and effective. But if your goal is to actually effect business change, you'll probably have better luck talking to a product person offline and having the idea come from them.
I agree with this to an extent but I feel you should always ask yourself 'why'. 'why is this feature valuable to the user?', 'why has this functionality been chosen over another?' etc.
At the end of the day the work you are doing has an impact on opportunity cost - if you at least internally question the work you are doing, you can potentially lower the opportunity cost and therefore help the business.
However, if you are part of a team of people who consider these things before the work gets to you, I agree it can be counter-productive to question it.
Re: questioning the work... I’ve found that there’s a certain threshold (that is surprisingly high, sometimes) where it’s easier and faster to just do something low value than it is to convince everyone else involved and reach alignment that it isn’t a valuable thing to do.
Regarding the todo list, I find that having a text-format todo list for the current day can provide structure and sense of accomplishment (especially at the end of the day). After completing a task I would put a checkmark emoji in front of it. Such a list can also bring me back on track after a lunch or a mentally-demanding task.
If I know that I won't be able to do a certain task today, I put it in advance for the next day's todo list. Not completed tasks will roll to the next day.
Plan for May 11:
- Contact Person 1 regarding Thing 1
- [checkmark] Review Pull Request 1 made by Person 2
- Implement Thing 1 from Feature 1
- Read Email 1 from Person 3
This is similar for me, except I integrate that into a daily log via a running google doc. Lists off the tickets I did, meetings, conversations and action items to follow up on. We have two week sprints, and sometimes around retro its really easy to forget things that happened the week before that would be useful to call out.
I guess this means "park facing down hill" as opposed to "park down the bottom of the hill" which is what I first thought it meant but doesn't make as much sense. lol
Learn your environment and learn it well (over time). Every single thing that seems like weird arcane magic invented by brilliant wizards is just code or words, and often not a lot of it. Go find that code and read it. Docs are good, but they are almost never complete enough, because truly exhaustive docs would simply be the code [1].
Similarly, with something like CSS or OAuth that have specs, blog/tutorial content often leaves out important details that will be relevant to you. If you get stuck, quit wasting your time with them, and go read the damn spec! If you're still stuck after that, go read the implementation.
Wherever you work, there is probably a page with like hundreds of ugly graphs of it. Learn what the graphs mean.
When you do this, a day will come where someone will say "Kanye, I've been banging my head all day on why users are sometimes seeing outdated results after updating their page". A little thing will happen in your head, and you'll say "That sounds like replication lag/a weird quirk of the CSS spec/a bug I noticed in the graphics code", and suddenly you'll discovered that you are the wizard.
I'm not saying to do this instead of doing those other things, or instead of whatever you're supposed to be doing. I'm saying do it as part of your practice of making software. You're probably gonna do it for like 40 years, so might as well make the investment.
Something that helps me is drawing/writing my own summary of what I just read or what a process is. For example: OAuth2 can be pretty confusing at first, but reading about the flow and drawing your own diagram helps me a lot to learn, as opposed to just reading. It also makes a nice reference for later, tailored to what I need most.
KEEP NOTES. Write everything down. Write down what you were thinking about, how you felt about decisions, things you tried for bug fixes. you never know when this stuff will be useful to you again.
I keep really basic markdown notes in a git repo, roughly one file per day with an ISO date name, but some things I give a separate file name (still with the iso date, but with some descriptive words) for things that might be good to re-access or are part of a series of similar notes (like a TIL file or a recurring meeting with the same person).
I do the same, using the Zettelkasten method. I've found it very useful. I write many small notes as markdown files and link them together, storing them in a git repository.
Yeah, that entire article is generated from the set of notes. There's one note for each subsection (down to the finest level of granularity) and it just extracts the article directly based on the links. I described this in a set of notes that end up here: https://github.com/alefore/weblog/blob/master/zettelkasten.m...
The images (e.g., distribution of back links per file, semilattice view, etc.) are also just generated programmatically from the repository of notes.
Here's another example of an article generated from my notes (this is way longer and is very incomplete, and probably contains a bunch of mistakes, but you may find it interesting as an example): https://github.com/alefore/weblog/blob/master/software-corre...
The first thing I do for every project is set up a development blog. Hugo + a nice free theme + Gitlab Pages + Netlify CMS. Everything is contained in Gitlab and I can keep the Pages site private and only accessible by members of the repo. It is less of a daily log and more exploring "issues" - basically anything I took time to research and figure out. I'm very adamant about recording details like dates, version numbers, context, code snippets, commands, and URLs of resources used. It all comes down to asking myself, if I see this issue again in the future, what information do I need to easily solve it without opening 40 browser tabs?
At first it takes some effort to circle back and write the post after you solved the issue, but after a while I found myself writing the post concurrent with writing my code. The Netlify CMS isn't the most beautiful UI but being able to just write markdown in a browser window eliminates having to keep a git repo locally. Organizing with categories and tags makes referencing published posts easy and if you want to get fancy you can integrate Algolia for search.
Honestly, having a project dev blog has saved me so much time since I don't need to keep every detail in my head anymore. Being able to go back and read about why certain decisions were made or how I overcame complicated problems is absolutely invaluable. And of course having this rich documentation while on-boarding new team members or passing the project off to someone else makes everything easier. Setting up a dev blog is 100% worth the minimal effort it takes.
I've been doing this for years, and even though basically no one reads my blog, it's nice to have a place where I put "polished" thoughts about my projects.
I love this. I have a grouping of pages in Notion for tech notes. But I really like the idea of scoping it to a project. I like that you don't need a local repo. I'll have to experiment with this. Have you automated anything for it? And do you have any patterns you use (eg. organizing the information, etc). Thanks!
There's really not much to automate. The Gitlab CI/CD config file to build Hugo and publish to Pages is always the same so I just re-use it as boilerplate. Likewise the Netlify config is usually the same as well, so once you've learned that syntax and structure it's not much work to put in place. I always like to peruse the latest Hugo free themes before I start and I usually end up using a new theme for each site anyway. The only real patterns in terms of organization is making sure I categorize and tag each post in a detailed manner. Most issues are related to a certain piece of the project or tech used, so when you know for example that your issue is related to the database then you can look at all posts categorized or tagged in a similar way to see if you already addressed the issue. I do find the most helpful step I took was making sure the post had a nice meaningful title with the proper keywords - that helps keep you from reading multiple posts until you find the one you need. Hugo and Gitlab's CI/CD are really great but the open source Netlify CMS is the key to making this setup so easy and remote only.
I tried to do this for my last project but ended up spending too much time making the blog and didn't get around to properly starting the project! Maybe instead of rolling my own site from scratch I'll use Hugo.
A similar idea I practice. After I solve a complex problem I usually leave a big comment at the top of the main file, explaining how I arrived at that solution, the caveats, how it works, and how it interacts with other files and dependencies.
If I need to, I also use Monodraw to produce ASCII character graphs and add them to the comment to explain concepts.
It helps to cement the solution in your brain and it's a system of record on how it was accomplished in case you forget.
I do this only for personal projects though. I don't think co-workers would appreciate an essay with exhibits at the top of a file, lol.
+1 for notes. I make really detailed notes in issue threads when I'm working on things. I recently started keeping a TILs GitHub repo and it's already proved useful. https://github.com/simonw/til
I want to take notes everyday, but I keep getting bogged down by how to organize those notes so I can "look back" efficiently. Essentially, I want to avoid taking so much notes that I'm reluctant to search for something because I haven't organized it well. Do you think the ISO date name method is working well for you?
It's mostly for sorting and context. I just do a straight up search (in my case in vscode) across all notes when I want to find notes about something particular.
Really the most important thing to do is just start taking notes. You can reorganize later if you need to.
note foo # creates a note in vim
note # lists notes in vim’s netrw
Since note file names are time stamped, they’re sorted and I don’t have to care about dupe names. It defaults to creating as .md but renaming the file afterwards doesn’t matter.
It’s so dead simple that I use it all the time instead of fantasising about the perfect note system.
I use org-mode in emacs for this! I usually keep notes on whatever task I am working on, and more general programming notes which I file into tagged headlines for easy reference later.
Yes. I wish I did this years ago. I’ve been doing it since January but there are so many things on my resume (and elsewhere!) that I’ve since forgotten and it’s annoying.
I use a combination of WorkFlowy and Notion for note-taking. WorkFlowy serves as my daily quick jotting medium. Notion I spend more time crafting and organizing, so that knowledge is more discoverable. Do you ever find yourself going through these ISO-dated notes? I would image re-finding something is difficult?
I do this too, and I use Sublime to generate HTML from Markdown, so at work I have hosted guides like commonly-used SQL queries for one of our projects or documentation on how I updated our OpenSUSE and built a specific version of mariaDB we needed. It's been pretty great.
This. I'm having some difficulty remembering details of systems I worked on and bugs/approaches I took at places I no longer work at. Going forward I'm going to document details about everything I do, a personal post mortem of sorts.
The biggest boost in productivity I've had in recent years is switching back to a statically typed language. TypeScript in my case. You can absolutely refactor and change course _so_ much quicker and with virtually no regressions at all. Just the other day we realized a feature I was building didn't handle an important case, and I had to really change the approach. Doing it in a JavaScript would have been dramatically harder.
The investment you have to make to really grok static typing is quite large. It took about 2 years for me to be really comfortable in TS. But you'll have to claw it out of my cold dead hands now.
Before switching to web dev, I used to work in Java, C#, C++, etc. But I've been away from static typing for about a decade now. I completely forgot how beneficial it is.
> The investment you have to make to really grok static typing is quite large. It took about 2 years for me to be really comfortable in TS.
That sounds a bit exaggerated from my experience. If anyone interested in TS got discouraged reading your comment, it took between like 2 days and 2 weeks for my team to be really comfortable in TS. If you already know JavaScript and if you already have some experience with statically-typed languages, TS is literally JS with static types.
Maybe my choice of the word "comfortable" was incorrect (although I would say "grok" was a spot on word choice in this context). It took me a good while to get the point where no matter what I encounter in TS, I feel pretty confident I can arrive at understanding it. The type system is very powerful and there are some deep rabbit holes you can find yourself in.
For example, if you can follow this blog post after only two weeks then you are much smarter than me: https://medium.com/free-code-camp/typescript-curry-ramda-typ.... This is a pretty extreme example, but sometimes you hit crazy types or crazy type errors like this.
But I would agree it doesn't take all that long to feel comfortable enough with TS to be productive with it.
Do you use ReasonML on the backend as well? Or just something like ReasonReact? I've been interested in Reason and still wondering the best way to work it into my NodeJS/React/Typescript stack.
I am not parent comment's author, but I use TS on the backend and I did look into ReasonML.
I love most of Reason's syntax and FP in general, but unfortunately, reason/bucklescript ecosystem just isn't mature enough for a serious project. I'd recommend against using it unless you are in some megacorp and you can afford to build general-purpose tooling.
Used to love Python. Now, unless it has type hints, not so much. Used to hate static types, or at least I thought I did. Really, I just hated C++ and Java. Go is Bae, trying to learn some rust.
Computers are rigid. The more rigid the lang you talk to them, the less you get bitten.
Untyped python, bash, javascript, just not worth it imho.
FYI: Perl has moderately strong typing, as well as forward references, an excellent debugger and probably the most powerful web framework, mojolicious.
If you're unhappy with how loose python and js are, you might find happiness with Perl.
TL,DR: Try F# or other "impure" ML languages if you want the security of types but you don't like actually typing the types.
I also used to love js / php because because I thought types make the code unnecessarily verbose. After some pet projects with a strongly typed lang (c#) I realized the peace of mind that using types gives. But I still maintain my opinion that types make the code too verbose.
But then I saw the "light" that is F# and I'm here to preach:
F# has types BUT you don't need to type them (most of the time) because it has type inference.
F# can be used in the .net ecosystem and most of the time it can replace C#.
F# can be compiled to js (although the interop is a bit messy).
F# is a better typed javascript than typescript.
F# is predominantly functional but you can also write code in the non-functional(ha!) style of your choice.
F# discourages you from coding mutable spaghetti
F# has decent tooling: VSCode + Ionide + dotnet + paket
F# has pipes |>
The bad parts:
F# is not a hard-core functional language like Haskell, but for me this is a positive
F# is a functional language which requires a change about how you think about a program.
F# doesn't have a huge community
F# doesn't give you a lot of job prospects
F# doesn't have a lot of 3rd party libraries (but you can use c# libraries in .net or js libraries with fable. You can also mostly port libraries from other functional languages)
F# requires you to think about the order of your declaration. Cyclic dependencies are harder to accomplish although this might be a good thing.
For the good parts, I would add the effortless type declarations to overcome “Primitive Obsession”, the immensely useful pattern matching to ensure you cover all edge cases and what I really found to be the killer feature: Type Providers to enable typed data wrangling in just a few lines.
That's what I meant with "unless it has type hints" but thanks for the link :)
Blue ocean python I pretty much start with hints and interface definitions, the logic follows. Helps me through writing the logic when I have autocompletion of methods.
I like Ruby for prototyping and quick throwaway scrips. I can write code really fast in it. Still my favorite language. But reading unfamiliar code or cooperating with others is hard. This is a good step forward in that direction.
Yeah, TS has one of the most powerful type systems I've come across, with its distributive conditional types and type narrowing logic. Recently I've been using type predicates to "dynamically" expand 'this' in specific clauses.
It's very fun to work with and allows for really safe code.
Same realization here, but from JavaScript to Rust. I'm still a bit more productive in JavaScript but I now regard it as technical debt to be rewritten in TypeScript/Rust as soon as possible.
What kind of systems are you developping with rust? Especially if you also do javascript? Genuinely curious as I also want to start working with some lower level languages.
This is really cool. I had a similar idea a couple of years ago to implement something similar to this using the three finger single tap command on the trackpad that currently brings up dictionary/siri knowledge/wikipedia etc. I see you have 4-finger tap implemented with an additional app which is really cool and probably better than the 3-finger native OS way since you can highlight/copy as well as build your own menu.
Probably an unpopular opinion but... Write your own text editor.
I've been working on my own (https://github.com/alefore/edge) as a side project and using it exclusively for about six years. I don't expect it to be very usable by other people (it's very customized for my workflows, I suppose; e.g., it's mostly useful for writing C++ and Markdown files) but, because I know it inside out (and I've invested in making it easy to customize through its extension language), it's very easy for me to adjust it to behave exactly the way I want it, which allows me to lower the friction for any operations I care about (and it abides very exactly to my principles/expectations for how an editor should behave; e.g. never block the main thread/UI while executing some operation such as opening a file). Because I don't have to use but a small set of computers directly (mostly my laptop and my workstation), this works well enough.
I don't know if overall it'll save me more time than it has taken me to implement it, but I do believe it allows me to move significantly faster than if I still used Vim or Emacs (or than most of my team mates), especially because it allows me to operate at a higher semantic level than otherwise, eliminating distractions from lower level details.
... and, I guess, it has been a lot of fun to implement it (and I've probably learned a bunch). I think it has played a role for me similar to that videogames have played for some of my friends (e.g., this weekend's challenge may be to generate visualizations for the logs I keep for every interaction I have with each file I've edited; implementing stuff like that feels similar to how in the past I felt about making progress in some videogames).
Genuine question: do you think that writing your own editor was a better investment than extensively customizing an existing editor like Vim or Emacs (even going so far as to change the keybinds and UI)?
I guess I'll never know for sure. I believe it has been worth it for me, as there are several things I can do now that I'm not sure whether I'd have been able to achieve by customizing them (unless you include "rewriting their source code extensively" as customizing them; but at that point I think I'd be roughly doing the same as what I did?).
I'll give three examples:
1. Being able to run
edge -view=all src/*.cc
This loads all those files, splits the screen, showing all matching files (up to a minimum area for each, with "Additional files: 498" (not shown) at the bottom), and lets me modify all files simultaneously (e.g., "search for a given regexp, make all cursors active, advance 5 words, delete to end of line, save all files, quit).
I just recorded an example of that (where I'm just renaming "OpenBuffer" to "Buffer"; you can trivially do that with Perl, but obviously you could do much more than just a simple regexp replace): https://asciinema.org/a/XNbNGL38kOrok2HO7zrarrQad
These are things that even heavy Emacs/Vim users would typically do through sed/awk/perl; I see that as a limitation in their editors (since you wouldn't use the same sed/awk/perl technique if you're just going to edit a single file).
2. I've supported multiple cursors (within one buffer) natively for a while (and I use it very often; for example, searching for a regular expression just leaves a cursor in each occurrence).
I guess I'd have been able to accomplish things like this, but I'm not sure of the quality of the results. In other words, I feel that it would have to rely on putting a lot of complexity in extensions and I'd guess that it would be too brittle and difficult to maintain. There are probably extensions for these things for Vim and Emacs, but I would be slightly worried that they may not integrate very well with other features and may brittle. But I don't really know.
3. I also got fed up that these editors would block on most operations (such as when you typed ":make" in Vim or when you opened a file from a networked file system); my editor never stops responding to user commands (rather, it simply visually indicates that it is executing something; perhaps you'll see side-effects as they occur). For example, here is how "make" works (you'll see me switching back and forth; most of the time you'll see the dots next to "make" (at the bottom line) moving, reflecting make's progress; in case it helps understand what's going on, I save the file, which causes "make" to be killed and restarted): https://asciinema.org/a/es4O4UdxPzB0vl7Tr88TKlq9N
I bet you can make Vim/Emacs operate this way (compile asynchronously, overlay errors with the files (as you can see around second 0:31); be able to nest commands with a pts within them, so that you can use them as you'd use tmux/screen). I'm not sure you make them load/save files asynchronously, never blocking?
Those are the examples.
When I started I was a heavy Vim user, but I got fed up of having to edit vim syntax, which I considered a, hmm, suboptimal programming language (yes, I'm aware that there are bindings for nearly every language under the sun, but still). I considered Lisp slightly preferable (and at the time I was still somewhat enamored with Scheme; I had been contributing some ~important modules to the Chicken Scheme implementation; these days I'll go to great lengths to avoid coding in languages that make static typing difficult, mostly because I don't think I'm smart enough to use them successfully for large/complex enough projects), but I was more into Vim than into Emacs. But I felt like it ought to be possible to do better than either. I felt that they suffered from carrying a lot of assumptions that were valid in the 90s (or earlier, perhaps 70s) but no longer applied, so I wanted to see how far one could go and experiment.
For example, as the user is typing, between each keystroke, something like ... hundreds of milliseconds pass. That's an incredibly long time for a computer! However, these editors mostly just sit idly, waiting for the next keystroke, not doing anything. My philosophy is completely different: burn as much CPU as you want, as long as you can give me something useful in return (and as long as you never stop responding). In other words, do whatever you can to maximize the value for the user.
You can see an example of the type of things I mean in the prompts in the above recordings:
- In the 1st recording (https://asciinema.org/a/XNbNGL38kOrok2HO7zrarrQad), around second 0:16, where I start typing a search regexp. As I type, the editor tells me things like "this would match 394 positions; in 32 buffers; and there's 2 search patterns in the search history that this matches".
- In the 2nd recording (https://asciinema.org/a/es4O4UdxPzB0vl7Tr88TKlq9N) around second 0:12, where I start typing a path (of a file to open). The editor scans the filesystem (asynchronously, obviously) and history log and tells me something like "you've typed `buffer_` so far; this matches 17 files (in all registered search paths) and 8 entries in this prompt's history".
(The key point is that all this functionality is asynchronous so it never blocks the user. If you type the next character as it is still scanning something, it just throws away those partial results (I'm somewhat simplifying; it's a bit smarter than that).)
You can probably achieve these things with extensions for Emacs/Vim, but I'd guess you'd still be somewhat limited by assumptions they make?
At the point where I'd be basically rewriting most of their source code ... I think it'd have been a significantly longer route (because I'd probably have had to care for a lot of additional things that are irrelevant to me).
Anyhow, to wrap up (sorry for the long rant!), this has been a great experience for me. I've learned a lot (e.g., I think I have more informed opinions about things like fuzz-testing, or the use of settable futures vs continuation-passing-style vs callback spaghetti) and I'm somewhat doubtful I would have been able to achieve so much through my own custom extensions for existing editors.
Those are some super cool screencasts, thank you for sharing! What are you showing on the right side gutter of your editor, visible at 20 seconds on https://asciinema.org/a/es4O4UdxPzB0vl7Tr88TKlq9N?
It looks like the function or loop scope? What are you using to generate/output that information? What languages does it work in? How hard is it to add support for new languages?
Aww, thank you for your reply, I'm very flattered. I usually get very skeptical/cynical comments when I mention to my coworkers and friends that I use my own text editor, so receiving comments like yours is very encouraging.
Yeah, what you see there corresponds to the syntax tree of the file being edited. You can see it fairly clearly in this screenshot (which has the advantage of showing how this really looks in an actual terminal, without some noise from asciinema): https://raw.githubusercontent.com/alefore/edge/master/screen...
In this screenshot, the very last column is just a scrollbar. Right next to it you see a representation of a subset of the syntax tree. It is a subset because I deliberately drop from it any subtrees that don't span more than a single line.
In the recording you also see something on the right of the scrollbar. The things shown on the left of the scrollbar correspond to the lines currently shown on the screen; the parts on the right are a representation of the tree for the entire file.
The syntax parsing is a bit dumb: just just based on things like parenthesis, comments, quotes, and brackets and such, not really aware of high semantic concepts (like classes or methods). For C++ (and other C like languages such as Java) this is implemented here: https://github.com/alefore/edge/blob/master/src/cpp_parse_tr... (if it seems too complex, part of it is because I cache previous states, as an optimization, since this runs every time you type a character into the file, refreshing the tree; I want this to work with very large source files, where most of the time you type a character the vast majority of the previous outputs can be reused).
I have two other parsers, for Markdown and diff (patches). It should be easy to add more parsers, but I want to clean up the API used by them.
Anyhow, you just have to generate a representation of the syntax tree (in this structure https://github.com/alefore/edge/blob/master/src/parse_tree.h). I've been considering integrating this with clang or vscode but haven't gotten around to it.
Once you have the ParseTree, the editor will display this information about the scopes (the logic that does that is a bit complex but lives here: https://github.com/alefore/edge/blob/master/src/buffer_metad..., function DrawTree, which returns a string with the characters for a given line).
I probably gave you more information than you wanted. :-P Thanks again for your comment.
This has been put to the extreme by Chuck Moore [1]
> During the 1990s, he used his own CAD software to design several custom VLSI chips, including the F21 processor with a network interface. More recently, he invented colorForth and ported his VLSI design tools to it
I'm so glad to see this comment. I recently got to the point where I created a text-viewer. Then I realized: This is the weak-sauce version of just admitting I need to write my own editor. My toolset is kind of weird conceptually so nobody else has created "my" editor yet. Haha :-) Thanks for sharing your experience.
Thank you for your response. I think most people here would disagree, but... I'd totally encourage you to go this route. It has been very fruitful for me.
Writing tools, and updating your dotfiles with helpers/shortcuts is absolutely useful.
I had a small collection of useful utilities become somewhat popular on github, even though I mostly published them so I had somewhere central to install from.
These days I've been reworking them into a single binary so that I can deploy them even more easily:
I agree. So many times I've blamed software for not doing a thing, then _actually_ read guides and documentation and find out it's me that's the idiot, not the software Devs. Over and over again.
I can't say enough good things about a multiple clipboard manager. Not only is it useful to copy/paste multiple items in sequence, but it's a game changer to be able to quickly retrieve a code snippet, or URL, or CLI command you used recently; and/or, to trivially stash one just in case you'll want it later. (Primary downside: using a computer without multiple clipboards feels broken. What do you mean it "forgot" the thing I copied five minutes ago?)
My weapon of choice these days is LaunchBar [0]. Despite the clipboard history feature being bolted on to otherwise unrelated functionality, I love the low-friction keyboard shortcut: Cmd-\, arrow-down to the chosen item, release Cmd to automatically paste.
The searchable clipboard history in alfred app set to three months has been a complete game changer for me.
Pushing things that I may or may not need at some point into the clipboard has become subconcious. The type to search is a dream and can even be used without really looking.
I have no need for any actice code snippet managment/notes since anything that matters stays in the three month buffer.
Nice! I've been looking for something like this. Do you use tmux? I wonder how it works with tmux copy mode... Semi-related to this, I wish there was a way to visualize your undo/redo tree. I hate that if I want to undo a change I made 10 changes ago, I can't (at least, I don't know how to). Something like a jump list in vim.
I used to have Glipper installed in Gnome, but it looks like that's been a dead project for the better part of a decade now. Does anyone know a modern equivalent? I did enjoy having a clipboard manager
I guess one potential downside is that I use `pass` to manage passwords and I don't terribly want my passwords all stuck in my clipboard history...
Yes! And please don't choose one that's too much powerful (usually means complexity) and bloated. On Windows over the decades I've tried a lot of them, and ATM, clipdiary is the best, because it does what I need (just keep a list of clip history and let me select and paste any of them handily) and not slow.
Write meaningful git commit messages, always. Even in stupid side projects that don't matter. Because then when you're at work, it will be second nature to write good messages, and your colleagues will absolutely appreciate it.
Nice! I totally agree with this! I've used https://www.conventionalcommits.org (initially from Angular, I think) and it has _really_ made me a better programmer. Worth pointing out that it also simplified code review a lot to make PRs less of a frictionpoint.
To add to that, rebase and generally "clean" your commits before pushing them out. A repo that has no "oops, typo" type commits in it is just so much more pleasant to navigate.
This. But I always emphasize the reasoning for doing so. I usually don't care about the git _log_, but I'm using git _blame_ very often. I want to know why exactly someone changed a piece of code, and "oh, typo" or "fix bug" don't help at all. At least include some ticket reference.
A couple others have mentioned shell tricks. One of my favorites is using Alt-. (in Bash and Zsh) to insert the last argument of the previous command. Press it multiple times and it will cycle through the last argument of all previous commands. It's great for when you want to, say, `ls thing` and then `vim thing`.
Yesterday I went looking for a similar key that would insert a copy of the last argument on the current line, like for when you want to copy or rename a file to a similar name in the same directory, as in `cp some/long/path/to/foo.txt some/long/path/to/foo2.txt`.
I couldn't find a command for this, so I made my own Zsh "widget" and bound it to Alt-/ so now I can type `cp some/long/path/to/foo.txt` and hit Alt-/ to prefill the second arg. I put the code for it up here: https://gist.github.com/dceddia/0a45a88dd61cbdf1be023d0a8f15...
Ooh, that's cool. You might be interested at z[1]. It kinda fills the same purpose as your trick, except it automatically figures out what are your most used directories by tracking your shell navigation or something.
I just looked up CDPATH and my understanding is that the difference between the two is that with CDPATH you are setting directories that you can quickly navigate to subdirectories of.
This is more of a quick link to directories themselves.
Maybe you could do something similar to the alias if you had my `~/.links` directory in your CDPATH, but then you'd still need to add the `-P`every time to get it to follow the symbolic link.
Not something I had ever thought of before honestly.
Yes, I love this one! I'll check out the widget. Sounds cool! I have the same problem (it's usually changing filenames, so I just use fzf twice, then modify the second filename).
There was one bash feature along the same lines of this. It was like bringing up the previous command in a shell script, so you could modify it freely, then run it. It was useful for long command, I think. Haven't tried learning it though. Just saw a couple wizard colleagues using it, and it seemed cool.
Don't push a commit with a new feature without tests and documentation. Even for your personal projects.
I started doing this a couple of years ago and it's hard to overstate the positive impact this has had on my macro productivity. I can come back to a project I haven't touched in six months and instantly start productively making improvements.
Tests and documentation really are for future you.
+1, good tests are by far the biggest long-term productivity win, in my experience worth much more than any other tooling or process improvement.
Unfortunately I've gone in the opposite direction as you. I recently moved from a company that had a good automated test culture, to one that has a culture of manually testing things, often in production. Everyone, even the senior engineers who've been around for a long time, regularly lands changes that cause prod incidents.
There's actually quite a good canary/rollback system in place to deal with this, so a lot of people think everything's fine. But they don't quite realize how much more efficient things are if you can just make a change and run a test suite, and quickly have confidence that you didn't break things.
Working for Eventbrite really helped we with this. I was there for over five years and got to see the transition from ad-hoc testing and an often broken master to every commit passing the test suite for most teams.
It got to the point where I could drop in on a project by a team in another country that I'd never encountered before, read their README, checkout their code, run the tests, make a change and submit a pull request.
The big thing for me in the past couple of years has been applying the same discipline to my own personal projects and finding that even there my overall productivity improved rather than being held back by the extra work invested in testing and documentation.
Absolutely! I've come to the same realization. Just haven't figured out my effective documentation rules. Difficult to experiment with what works when the feedback loop is on the same magnitude of how long it takes you to forget (eg. 6 months). Best proxy I can come up with is "how can I communicate this to someone who has no context? What would they want to know?"
I usually comment with the same thing on these tip threads. And that is learn to rebind your keyboard. Make each key programmable modifier. Open apps, run scripts/macros instantly. My Karabiner config:
I go as far as binding keys to instantly type `console.log()` or `fmt.Println()` or the other language equivalent with my dot modifier key. Plus typing things like `Thank you` with dot+spacebar press.
One thing I worry about with these extremely customized keyboard configs and such is that once you get used to them, you will have a hard time using any other computer when you suddenly need to. Maybe this isn't such a big problem with physical keyboards, but can be with e.g. text editor configs, which you can't take along if you need to SSH into many servers during a workday and work with whatever is installed there.
I always find this argument nonsense. Most people will be on their computer for 99% of the time and sacrificing for the rest of 1% doesn't make sense.
I had a colleague who had been using vim without customization because of that reason and he's always on his computer but I find the reasoning very odd.
as a caps-to-ctrl remapper, I can't imagine how crippling it would be to have serious remaps or Dvorak or something and try to use someone else's keyboard. Doesn't seem worth it.
Maintaining a Wiki or any sort of extensive note taking operation requires a lot of dedication and I guess practice? Do you have any tips for someone who wants to start out? Is there a smaller step that one could take initially?
I don't use a wiki but instead use Bear (https://bear.app) which gives cross-referential tags and a slick sorted tree of tags. I tag every article ('note') with a set of tags at the top.
#d/2020/05/08 #fam/sis #fam/gifts — notes around a conversation I had with my sister today about potential gifts .
#d/2020/05/05 #recipes/dinner #recipes/c/italian — notes a ravioli recipe I made on 5/5. If I later make it for a group dinner, I'll tag it, too, so I can track when I made things for who and collect longitudinal feedback and recipe changes.
#todo/today is things that need to get done today; things that don't will get moved to other notes.
In case you want to generate a graph of your notes (via graphviz for now), I wrote a project recently (I'm working on a D3 version of the graph "at the moment" although it won't happen until at least tomorrow, today is off-programming day): https://github.com/rberenguel/bear-note-graph
Holy shit... that wiki and alfred workflow is INSANE. I am so jealous. Do you find that you're able to discover what you're looking for pretty instantaneously? Any general tips for organizing (sorry if you cover this in the wiki, haven't dived into it yet, but i'm going to!)? also, huge fan of keyboard remapping. I use karabiner, kinesis advantage, and a couple kinesis foot pedals (had some rsi issues, but all this opened my mind to the possibility of experimenting with key mappings and input movements).
This is a bit of a departure from the question you asked, but I'm always amazed how much people tend to hyper-tune their dev environments, when in reality, most of my time is spent thinking how not to write code. I get just as much work done on a completely vanilla macbook with VS code installed on it as my editor-obsessed peers with obsessively optimized setups can.
How much actual code are you writing? I'd be really concerned with myself if I am putting out so much code that I need to be concerned with my text input speed.
Your tooling isn't just about writing code, it's also about reading, analysis, debugging, research. You think about the code, have an idea, need to look up how something works, think some more, have this constant back and forth. Optimizing your environment minimizes the friction, and allows you to immediately get answers as soon as you have a thought about something, so you don't lose your train of thought.
You also want to be able to run quick experiments, same idea, get the answer to a question you're thinking about as fast as you think of it. And then once you actually know what you want to do, don't you want to have as smooth an experience executing your fait accompli? The primary focus is minimizing friction, and speed is just the observable side effect.
And then of course you can't ignore the natural human enjoyment that comes from customizing things to exactly how you like them.
I spent quite as few years as a contractor and I quickly discovered the best way to be effective was to learn how to be effective with everything in the default settings. At every new place I went I could just jump right in. No need for any time spent fiddling with settings or tweaking and installing pluggins.
Although a slight counterpoint is that I also used to have my git settings file stored in a github gist so I could just grab my latest one and every new environment would have the same set of aliases I was used to.
I donno. I use IntelliJ and when I pair-program with someone who knows how to use its features vs someone who treats it like a basic text-editor/compiler, is night and day difference.
If I am 'riding shotgun' and say stuff like "look up that definition" or "find all the implementations of that method" or "go to ABC" and they just look at me like a deer in the headlights, or start clicking through the file tree, or do "find text in project", then they are going WAY slower than they could be going.
So if "obsessively optimized" includes learning your tools, then it returns rewards.
Lines of code are the wrong metric and turning dev environments isn't going to get you to type faster. If you're happy with VS code on OS X with zero customization, more power to you :) The "obsessively optimized setup" isn't the results of a single week-long binge, which may lessen how obsessive that seems. A setting here or there, after a few years, results in a comfortable dev environment. The comfort of this environment extends the ability to make code happen.
Coding is, as you've noted, rarely about typing speed. The larger a program gets, the longer the files get, the more spread out across different files and directories it gets. (Consider yourself very lucky if the programs you work on can fit entirely inside your working memory! Others have bigger programs to work on, or differently shaped working memory) Being able to find "that thing you saw that one time" is paramount. How you find it, is an entirely personal choice, but being able to find it, quickly, and without breaking out of flow state, is the goal. Whether it's by learning VS code (which has a expansive collection of settings and extensions to "obsessively" tune), or by installing ag. (And yes, for the obsessive, the merits of ag vs ripgrep/other is a fun rabbit hole to fall into.)
Still, between from when you start coding, to when you share the diff with anybody, coding is an intensely private activity (that some choose to share with others). As such, whatever works. If trying to open a file while traversing a tree view is distracting enough that it takes couple tries, and you don't mind taking a couple times, great. If I've got my dev env setup to so I can open the file from the command line where I've already got the filename, that's cool too. If I want to learn VS code's command bar syntax so I can open it from there, that also works. And if you can open that file first try, while simultaneously navigating a stream of code, good for you.
End of the day, comparing "just as much work done" with two independent variables (the configuration, and the person using the setup) and no control, is as foolish as managers trying to quantify programmer output by counting lines of code.
For the most part I agree with you, but it got to the point at work where I could regularly type faster than eclipse (usually some shitty background tasks messing things up but still) this actually slowed me down as well as making me angry, which is probably the way to make me least productive and ranty. Like, how can this be the case that in 2020 I can type faster than a computer displays text in a professional IDE. Jesus christ. Oooo calm. Using sublime text now and just using eclipse for the debugger. Calm. Calm.
It isn't so much the total LOC that I write (and delete) in the process of making a change so much as it is the friction and mental overhead of the change. The more I can focus on making the semantic change that I care about, the better.
Admittedly my vim setup doesn't even have YCM because I don't care enough to do it. But there are other minor things (swapping : and ; in control mode, mapping jk as <ESC>) that I've done that I hate working without now.
The REST Client VSCode extension[0] let's you send HTTP request from any file open in the editor. I use it to call the Cloudflare API to purge the cache of my site after I make changes. It's also really useful to test new APIs or quickly create a bunch of file/requests for the current project.
I use PHP on my landing page[1] so I can make changes and publish them immediately without any build step. This way, whenever I see something wrong with my site or have a new idea, I just open the editor, change the file, save and push all changed files through FTP to the server using the VSCode SFTP extension[2]. This means that I can instantly publish changes from VSCode without running any command, just by using shortcuts. This way I can make the changes immediately instead of having to create reminder tickets.
Automate the tasks you do often. For example, I had to download invoices from Envato every quarter, which was annoying because they don't have a button to export the PDFs of all invoices. I created a puppeteer script (headless Chrome) to download the CSV with the list of invoices, go to the URL of each of those invoices, print the file as PDF and save it into the correct folder. The automation part can also be just creating a faster build process, unit tests or even subscribing to notifications/alerts for a social site for the content you are interested instead of checking everytime if there's anything you might like.
I've just learned about this extension today from a Twitch stream. First I thought: "Nice, another great IntelliJ IDEA feature got ported to VSCode" (I think the syntax is exactly the same). But seeing the VSCode extension was first released in 2016, it looks as if JetBrains actually copied this one. Always nice to see how IDEs/editors inspire each other.
The REST client extension is so awesome, ever since I figured out how to use it I've been trying to convince everyone I know to use it instead of Postman, cURL, etc. It's just so easy to save common requests to a file and have everything in my editor.
Same for IntelliJ-based editor! It's incredibly useful to store the .http files inside git repos, an have it tracked alongside with code. It's similar to Postman or Insomnia, but tracked (and synced), and built-in
My usecase is that I often have queries for debugging purposes for API's that I need to send, so in my repos I keep a requests.http file for quick-access.
I also put them in repos for projects I publish (like boilerplate/demo API's) along with a screenshot and a note + link to the extension in the README because you have a working spec for API requests, and so do if/when I ever revisit it.
Tiny extension saves a ton of time and super useful.
This is awesome. I usually just have a scratch file that uses an http library (eg. requests for python) and log the response. Do you prefer this over something like postman?
Yes, I only use this now. I used postman for a while, but I remember last time I tried it, it was way too complex, had to create projects or accounts (or something like that) to make a simple request.
I now only use this VSCode extension, I have some .http files saved in my project with requests I do often. If I am trying an API online, I just copy-paste their cURL examples and run them, it's useful because on Windows there is no cURL cmd preinstalled. It's also nice that you get code highlighting by default in VSCode for the .http files.
Another big advantage is that you don't have to open a new console or application. You just do CTRL+N to open a new file and ALT+CTR+R to send the request. So you can do CTRL+N, write google.com, ALT+CTR+R and you can see the source-code of google.com in a VSCode file.
Having the request in a file means you can easily change it, save it or version it.
Set up a "scratchpad" repo for small experiments / learning, and phrase your learning as tests.
Not only does it get you practice writing code, you have evidence that what you think is actually true. And/or you can demonstrate it's not, when you learn otherwise. And if you push it somewhere, you can link people to it trivially (great for questions / teaching / bug reports).
It's pretty much guaranteed to be faster to hop in / build / run than your day-to-day repository (often by a few orders of magnitude), and testing some things forces you to learn some new tricks (concurrency tests / bug demonstrations are particularly complex). And personally I also recommend setting up an auto-rerun-tests loop (I prefer `entr`), which gets you very nearly repl-level feedback in any language. You can interactively explore to learn basically anything in any language.
Love this idea! I generally just have scratch files in whatever codebase I'm working on, but I like the idea of having a centralized place to keep these. Any chance yours is public? I'd love to see an example of how you organized the repo, what type of stuff your testing, any boilerplate setup/tooling you used to make it more effective (mocking library to stub out things like network calls or complex objects, test runners, etc)
No, just one internal to work (which I share pretty heavily), though I should set a public one up. I've got a few bad ideas fleshed out enough to realize why they're bad, and a few multi-million-iteration things to test the fairness of go channels / cancellation strategies / etc just because it seemed like fun. But the repo has slowly grown to include things like demonstrating issues in books[1].
Boilerplate setup for me is typically:
- a one-page Makefile for laziness. I try very hard to avoid automating what I don't fully understand, and large / complex makefiles are typically inscrutable for readers. usually I just have "build", "test", and "live tests", little else.
- a small script to use `entr` to run tests in the last directory changed (the /_ arg)
- ... that's about it. maybe a .editorconfig or .envrc? I avoid lang-specific runners because they're all different and tend to have complex relationships - if you can't run your test within a second or three, find a different way. break apart dependencies to benefit from build caches. leverage make. etc.
If I wrote experiments that used networking libs, I might have common stubs, but none currently :) I'd be sorta tempted to write fakes tho (probably multi-process), and try to integrate them into tests, in part because that's often hard for no apparent good reason. Maybe I'd find out why, or maybe I'd end up with a demonstration that there isn't a good reason.
I consider scripts to be functional documentation, so I write a lot of them, usually shell scripts but Ruby and Python, too. Any task that I might want to perform again, or tasks for someone else, gets a script if possible. All of my repos have a /scripts directory.
It can take some work to write and debug but in the end when it's working you know you have correct "documentation". Text and wiki documentation may look right but you can't be sure. Just today someone on my team was trying to create a particular build on Jenkins but was using the wrong build parameters, which were copied from wiki; the previous owner of the wiki had neglected a few changes. But the script worked!
I've found recently that Makefiles are my best friend for this kind of thing, even if it's just wrapping existing commands with the right arguments and conventions for the project.
Being able to do a simple `make build`, instead of remembering all the right command syntax, repositories etc is so much nicer that having to trawl through a README in a couple of months time for how I was using this thing
+1 I've been using makefiles more and more. And the nice thing is, if you don't have hidden dependencies you can just do a make -jn and get parallelism.
I also like having a script per project that will get auto loaded when you cd in. An example from a current project:
#my vimrc knows what to do with this
export VIM_TAGFILES="$PWD/build/tags,$HOME/.local/var/tags/system,$HOME/.local/var/tags/glib-2.0"
export CFLAGS=' -g -O0 '
function autobuild {
(mkdir -p build
cd build
while true; do
make
make deploy
inotifywait -qr -e close_write,delete ../cgi ../test ../Makefile.am ../migrate.sql;
done)
}
function watch {
tail -f build/app.log build/deploy/app.log
}
Over time they'll end up with a bunch of other random things. I've tried abstracting a lot of these to their own scripts but every project proves just different enough to make that more complicated. And most of them are personalized enough that they don't go into the project itself.
Another "super power" for shell scripts is writing completion scripts (https://iridakos.com/programming/2018/03/01/bash-programmabl...). I've had some insanely useful one's do things like look up database values to mimic a large complicated third party product, going through it's UI inputting repetitive info, copying files around, etc was reduced to fake-tool<tab><tab><tab><tab>.
You may find prm useful for this purpose, it lets you create projects with start/stop scripts that get executed when you do `prm start project` or `prm stop project`. I use it to quickly switch to the right directories, open editors, services, etc. needed by a project.
`direnv` will automatically activate and deactivate a project-specific shell env for you when you change directories.
I like to put scripts for the project in a `bin/` directory then use direnv to put the project's `bin/` on PATH.
I've leaned more and more towards mirroring the *nix directory structure in my projects. `etc/` and `bin/` are two top-level folders in a lot of my projects at work.
Whenever I onboard to a new codebase, or try to learn more about an open-source project, I typically create personal notes that outline the structure and high-level “markers” that I want to remember and easily reference later.
Unfortunately those notes are disconnected from the codebase itself, and therefore, not as immersive/rich as I wanted them to be. So I created a tool that allows me to annotate codebases with my own walkthroughs/breadcrumbs/ToCs and then play them back at any time, directly within my editor. This workflow has provided a pretty nice productivity boost for me thus far, and has the benefit of enabling me to share these walkthroughs with others as well: https://github.com/vsls-contrib/codetour.
To me, the key for being productivity is understand throughly the "tools" we use: IDE, languages, libraries, framework cli, os, shell,... Read the implementation, the docs, the issue tracker, even the git history if you have time.
When I started to learn vim, I tried many popular distribution without understand each of the plugins they included. And I almost gave up learning vim. It's until I start to read the vim manual, learn the key strokes one by one and then building my own distribution, that's when I really know how to use vim. Even so, I still learn many great things from books like Practical Vim by Drew Neil (on Tmux, there's Tmux 2: Productive Mouse-Free Development)
I think you might already know that, but still shooting here for some quick tips. I don't have any shortcut, just that motto.
Love this! I found that the more experienced I became as a dev, the more I gravitated toward the less-removed sources of truth (eg. man pages instead of stack overflow, source code instead of blog articles, etc). Haven't started reading issue trackers, but huge fan of traversing a git history (assuming it's clean and atomic).
I set up automation in my development to run quick tools like code formatting, linting, and unit tests together.
I'll use tmux to split my terminal into an editor window on the right, and on the left, I'll have a tool like watchman or entr watching for changes to my code, and running those tools e.g.
find . -name "*py" | entr make continuous
I also have some conventions about my tmux windows. Window 0 is where my music client lives in pane 0, and usually I'll run things like homebrew updates in other panes there.
I have a tmux keybinding that sends the "Next song" key to window zero pane zero. No matter where else I am in that terminal session, that keybinding skips to the next song.
I have another pair of tmux keys that activate/deactivate a particular openconnect vpn connection in a new pane, with a pause so i can see it complete, then it goes away. Part of my tmux status line is a single character - a C or a D, that indicates that vpn connection status as connected or disconnected.
I love entr! It's been part of literally every development environment I setup since I discovered it.
Re tmux: I have a function in my fish config (used to be zsh) that runs when I changed directories; if there is a `.tmux` script in the new directory, the function looks for a tmux session by the name of the directory, attaching if it exists, and, if it does not, creating a new session by running `.tmux`, which creates a handful of windows with my file watchers, build runners, container managers, etc etc. So all I have to do to boot up a project environment is move into the src directory, and poof — all the scripts that build source, run containers, and restart things on various config file changes get booted up immediately.
- Your eyes are the only broadband connection into the brain, and they have millenia of pattern recognition training built into them. When solving problems like debugging, refactoring or data preparation, consider arranging information in a way that takes advantage of this. When it's in the right shape, irregularities or patterns will often just jump at you.
- Relatedly, your eyes are also extremely well-adapted to detecting movement and optical flow. Do these two images/texts/logs/terminal outputs/etc. differ? Do they differ if you ignore the hue/time stamps/pwd? Alt-tab back and forth between them and you will notice. (Beware of fancy effects or menus covering it though - it has to be an instant switch from one to the other).
Oddly enough I can't use IDE's. They simply throw way too much information on the screen, and my brain gets overloaded. So I have to sheepishly confess that I do my programming in IDLE. There, I said it.
But I think a take-home lesson is that the "right" visual environment is a function of your eyes and your brain, and might not necessarily be the most sophisticated environment.
> They simply throw way too much information on the screen
You can usually just hide them.
I hide everything but the file explorer, the code itself and recently I turned on line numbers but no tool bar or task bar or that distracting mini map which I feel it serves no purpose than being cool for some people.
Excellent question. Actually my cube mate uses vscode and is happy with it, and the latest incarnations of WinPython come with vscode built in. So maybe I owe it to myself.
For now, my "IDE" is Idle on one side and Stackoverflow on the other. ;-)
I spent a short time trying to find a pretty diff tool, but couldn't really find one to my liking. So just ended up using a bash function that did some decorating on the diff output.
This allows you to (i) distribute your work tasks intuitively across space; (ii) completely zap out all distractions within a VR headset; (iii) completely eliminate the mouse and work 100% on a keyboard (VR Desktop allows you to click and browse with eye gaze).
It takes a bit of practice to learn the keyboard shorcuts (spawning terminals, moving windows, etc with keyboard shortcuts), but once you learn it, it can feel very immersive/productive.
I was imagining something like this for fully immersive computer work. Especially for development and research.
It can give you infinite screens.
One better idea, would be to tie in VR with some front cameras, and then overlay your windows and documents at specific spots in your room.
Like have your calendar on the wall. It becomes literally, a wall calendar.
Then a todo list on your desk. Some encyclopedias on your bookshelf. Etc.
This way, your working documents becomes grounded in your actual physical reality. And this gives you memory, of where that document was.
It’s like how my desk may be messy, but to me it’s perfectly organized. I know exactly where that receipt was placed, in that pile of papers in the far corner of the table.
Being able to recognize hand movements is also a plus, as this will make it easier to interact and manipulate your virtual objects.
Learning keyboard shortcuts in my most used apps has proven very effective. I used to have a bunch of cheatsheets printed but now use CheatSheet for OSX - https://mediaatelier.com/CheatSheet/ and CheatKeys when on windows - https://cheatkeys.com
Save all bash history into a separate file. Even with a high limit, the default bash history functionality has bugs that will destroy your history now and then. With the following .bashrc settings I have commandline history going back a decade:
Contrarian opinion, but often doing things manually rather than automated will save you time. Writing/debugging a script is fun, but more often than not it would have been faster to do things the dumb way. If a task is risky or repetitive daily, then sure, but often developers confuse "writing a script because too lazy to manually do the work" vs "get the task done as fast as possible".
Another contrarian opinion, but as much as configuring and tweaking conf files are fun, it's still faster in the long run to be highly efficient on a new system as fast as possible. It's also convenient to use standard tools to be effective on someone's else computer. For instance, it's much better if you can be productive by changing 2-3 keys in macOS settings instead of being forced to use a fancy keyboard layout and customize shortcuts for everything.
Respectfully, I've never understood this viewpoint. If the configuration options are importable from some external repository (as the vast majority of dotfiles are), "as fast as possible" simply becomes "how quickly can I install git (or my VCS of choice) and pull down my configuration options?"
For keyboards and other physical hardware, sure, you can't remap on the fly. But I'd wager the the vast majority of productivity-customizations that people implement are software-level.
As always, it depends on what you do. On your machines, sure. On customer machines, no, quickly installing a VCS and pulling some repository is often not an option.
People should realize that any kind of system or process comes at the cost of improvisation, which is always faster and lower overhead. Doing literally anything other than doing things as they come has a cost. The gamble you're making when you write documentation, automate a test, or even write code to begin with is that the investment in doing that thing will pay dividends sometime in the future.
1. Learn a good plain text editor really well. Learn the shortcuts. Learn how to navigate text (skipping words, end/home, etc.)
2. Start using REPLs - saves a huge amount of time when trying to figure out small quips of logic. REPLs are extremely useful whether you’re an artist expert or a beginner.
3. Learn how to setup and use debugging tools. These are often a PIA to get properly working, but you’ll thank yourself later. For example I do a lot of Ruby/Rails and rspec work. Enabling byebug and leaning how to run rspec feature tests with the browser in non-headless mode can solve some really tough bugs.
4. Learn regexes (POSIX). They’re used everywhere and have pretty much identical implementations and syntax between languages; they can help you do anything from performing complex searches to parsing logs to writing your own search engines.
5. macOS: Cmd+` (backtick) lets you cycle between windows of the same application, e.g., all of your Chrome windows. This was a game changer for me. I still see a ton of people not even using alt-tab or using that awful expose crap.
6. Keeping your system “up”: Caffeine for Windows, Insomniax for macOS. One of my biggest annoyances is systems that automatically lock, sleep, or time out, wasting my time and mental energy on logging back in or waking my computer.
7. Taking more breaks and procrastinating. Yes! Procrastinating does wonders for me. Not sure why, but some of my best work is done after goofing off for hours. It’s probably counterintuitive to most of you, but surprisingly I can get more done by spending less time forcing myself to do it.
While I suspect anyone reading this is going to detect at least a small hint of sarcasm, its true. Minimize distractions like this and unsurprisingly, productivity goes up.
My biggest one, to add, is having set time for handling administrative tasks, like Email etc. If you can (and I realize all people have caveats), just set a block of time in the morning and in the afternoon to answer email, and otherwise ignore it unless its been flagged a certain way.
This is what I do
first 30 minutes of my day, I do email catch up before my stand up. After the stand up I take about 15 minutes to update the KanBan, and then thats it for my admin tasks at that point. I may check email again (well I almost always) about half an hour before I leave for the day. The only exception is specific flags I have set and those will always notify.
I only have only one real tip for you: learn to get in the zone.
When I am in the zone I have an intense desire to create a beautiful masterpiece in my chosen art and revel in its elegance and beauty. When I am in the zone outside distractions are eliminated and my mind is fully focused on the task at hand. When I am truly in the zone matters such as health, family, children, work and (when I was younger) even hunger or the need to sleep disappear; all that remains is the task. My brain has maximum cycles to iterate again and again until I achieve greater understanding and/or arrive at a solution.
I do not have very good general tips on how to get in the zone. I have some things that work for me, although not always. They include:
- Real-world isolation: These days where everyone is at home, I go to a corner in my garden where my family knows not to bother me.
- Virtual-world isolation: No emails. No browsing. No notifications of any kind. Phone on silent. Absolutely no Slack, Teams, etc!
- Friction-free programming environment: If the build takes 20 minutes, then it will be hard to remain in the zone. If the Mac has stuck keys get them fixed or buy a laptop from a different brand.
- Headphones with some favorite music: For programming I like trance, the more repetitive the better. I often use this youtube channel: https://www.youtube.com/user/MrLemilica2
- Review previous work on the subject: This acts as a brain warm up for me and usually puts me in the right mood.
- Some times work better than others: When I was younger it was far easier to get in the zone at night. These days I do not have that luxury because of family.
- Work on a problem that excites you: It is much easier to get in the zone when the subject matter interests you. It is harder (although still possible) when the subject is less interesting.
- No context switching: Choose a task and stick to it for at least a few hours.
- Get the non-enjoyable tasks out of the way: (This may not work for everyone.) I like to complete tasks that I do not enjoy (e.g. administrative or legal) first. I can then continue my work care-free instead of having this niggling thought in the back of my mind: "do not forget to respond to the lawyer".
It'll provide similar benefits as writing your own text editor (which was suggested here).
It can also provide type hints and type checks for dynamically typed languages, thereby making them more like statically typed languages (also suggested in here).
Plus a good IDE will speed up searching as well as index the source code to make things click-able, thereby making large code bases or dependencies easier to navigate (also suggested in here).
So basically you can get a large chunk of the improvements that others have suggested in this thread simply by purchasing a good IDE :)
For a number of years I used the jetbrains IDE's (at alternating times I used phpstorm, webstorm and intellij), commonly perceived as best in class. Then I switched employers to a place that had standardized on VS code, so I moved over as well.
I have to say that while it was hard to get used to in the beginning, it is surprising how good VS code is. I get 80% of the featureset of jetbrains' products, for free. Specifically, all of the features you mentioned VS code does just as well when doing web dev.
The only thing I'm missing is the framework-specific automation helping you easily set up projects, builds, unit tests, and so on. VS code can automate those things, but there is more googling involved.
I think I've tried most java ides and they all suffer from the same issue which is that they're slow, unwieldy, and the benifit they bring (apart from debugging) are mostly not benifit at all. Learning how to write correct code without and IDE auto fixing everything you get wrong with a right click has been very valuable to me. Learning my build tool outside of crappy IDE integration has been an education too. Basically getting rid of my IDE has made me a MUCH better developer although I admit they do powerful and useful things.
Have a sharpened pencil and stack of paper handy always. I can't say how many times it's saved me a lot of trouble programming some behavior because I was able to draw a picture of what I wanted it to do.
I am opposite of that. I can't stand paper and pen/pencils, so whenever I need to visualize something it's Paint time. And if I need to note something down it's Notepad time (well, Notepad++ lately).
Automate repetitive tasks that can be automated. A task may take ten minutes. Automating it may take hours, and thus seems too expensive.
If this task is going to be done many times, then the automation WILL pay off. Running the automated task might take less time than the manual steps. And you can be doing something else.
The more of these kinds of tasks that you can automate, the bigger of a lever it becomes.
In my experience, that is the unusual case, not the ordinary case.
Further, when that would happen, it is typically obvious that some bigger change has occurred which required the automated procedure to be updated and tested.
Learning basic *nix commands like sed, awk, and grep has helped me so much. I still frequently google/manpage them, but I often find myself thinking "I can do this with X in just a few seconds!"
being medium-grade conversant in regular expressions is right up there, too. You might not need to do lookaheads, but if you can write a regex search in vscode that finds and replaces something across a hundred files correctly, you've just saved yourself a ton of tedious manual labor.
I keep two diaries, both MS Word documents. In one, I write summaries of what I read about -- papers, blog posts, books, and articles that interest me, along with the google search term, authors, URL, quotable quotes etc. In the other, I put together a free-form running log of projects I am working on, along with feature ideas, benchmark results (and the parameters used to get it),etc.
Why MS Word, you ask?
I am ordinarily an avowed plaintext person, and I have tried everything from various wiki formats, org-mode, markdown of every shape and stripe (even bought a few markdown editors), but they haven't worked for me. The reason is lack of WYSIWIG
WYSIWIG is a huge efficiency booster. Inline images, doodles of my own, colors, fonts, tables, inline math, greek and other symbols, all of it is right there in the original (not in a separate render), and is searchable.
Having all of it in one file is not really a problem performance-wise; I have had one file for the last 8 years. It is easy to back up and the chronology is clear, and it is easy to generate a table of contents.
Finally, having it on a google drive means that it is available/convertible to google doc if I need to access it from elsewhere. I do not rely on google docs to be the primary copy, because i don't want to rely on a working internet connection all the time. Offline works better for me.
I have not. The feature set looks impressive. However, I prefer my approach because everything is searchable text in temporal order. I don't wish to impose any further order on it, not just because it takes time, but also because it forces a particular hierarchy ahead of time. I often have mutual references to other articles, for example, or overlapping clusters of references.
No, any word processor should do. MS Word is quite frictionless because its document format is a defacto standard.
Of the word processors, I like Apple Pages a lot, and its inline math typesetting supports LaTeX. But it didn't at one time, and I already had a ton of stuff in my Word diary.
I did a quick trial run, and I love it. I'd pay for it. I tried drawing quick sketches on an iPad and pasting it, I tried latex math formulae inside tables; they worked flawlessly. The UI is gorgeous.
Lists don't seem to work inside tables though, and most of the keyboard shortcuts don't work for me.
In any case, for the theme at hand, I'd still go for a word processor and a single file combo. It turns out I need more primitives to organise my writing than even typora's markdown gives me, such as colors, formatted paras or lists within tables, tab stops etc. And most importantly, Word has evolved to support really large files, since I have had basically a single diary for many many years.
That said, I will try out typora in earnest for documentation and other things.
I've worked with many programmers who don't know how to touch type. Anyone who uses a keyboard for more than an hour a day should learn to. It only takes about 2 weeks to learn all the keys, but it's a skill that'll save you thousands of hours throughout your entire career.
It completely baffles me that some developers haven't learned this basic skill. I've heard the argument that thinking is more important than typing. I counter that by saying if you learn to touch type, then it's easier to get into "flow" and your resistance to writing code drops pretty quickly.
What first came to mind for me is running. It’s very difficult for me to sit still for long periods of time, and so I end up switching tasks often. If I do a good run, I can stay still and focused for longer.
I believe longer contiguous periods of time developing are very helpful.
I just recently noticed this too. I have a bit of ADHD and I've found it helps me follow meeting discussions and such. Can't quite explain it because being tired should make focusing harder (and true, I'm probably a slightly less effective coder than normal right after a run, just because of tiredness). But for whatever reason I've noticed it's much easier to follow other people's lines of thought and catch the important details after a good workout.
Edit: upon further reflection, I think the big difference is it's a few-hour boost in self confidence. So it's much easier to follow conversations and ask probing questions, rather than being afraid to ask things for fear of looking foolish.
It just works. You don’t have to use an image like Hyper-V. It’s up and running in less than three seconds. You can configure permissions on a per-container basis (with a lot of work).
Only bad thing is that MS doesn’t seem to realize what an incredibly useful tool they’ve built and don’t seem to be actively developing for it.
If you have to look something up or solve something and it takes more than 2 minutes then write the solution down in your notes. Lots of times you'll need that again and you won't have to search for it again.
I am a bit conflicted about the note-taking part. If you keep writing notes, at some point you will have to search through all of them to find what you need. So you not only spend the time to write the note, you still have search for the solution and there's the risk of the solution becoming outdated (eg. maybe the problem now has a better solution due to new versions of software)
Shorten your feedback loop by using a continuous unit test or code runner.
For JavaScript and it's variants I use WallabyJS (unit tests) and QuokkaJS (code).
For C# there's nCrunch.
For Python there's PyCrunch.
And for Java and it's variants you can find InfinitTest
Great thread, thank you. Already seen a few things that I will look into.
For me, the transition from Bash to Zsh has been a huge efficiency boost. Mainly because of some great plugins for Zsh, such as z, zsh-peco-history (better history search), zsh-autosuggestions, and zsh-syntax-highlighting.
Make aliases for your most common commands. Someone in a similar thread posted a script for parsing your history to find them out. For me they were all git subcommands.
Write a script the second time you have to do something.
On a macro level:
Don't work on doomed projects.
Don't update tools in the middle of a project unless the current version is a roadblock.
I have such a simple setup but it's not for everyone. It also fails miserably if I'm working in an environment where I can't do almost everything I need to do from the command-line.
1) VIM with almost no special .vimrc
2) gnu screen (10 screens initialized via .screenrc)
- hotkeys Ctrl-Left (prev), Ctrl-Right (next)
- I tweak my $PS1 prompt so it will display which $WINDOW I'm on.
3) Not a tool, but a tip. Maintain a TODO list. I just open a text file and put some simple separators between projects I'm working simultaneously. But I go out of my way to ensure it's got everything I'm going to "DO". If I think of something, even if I'm not sure if I'll actually want to do it, I don't rely on my memory for ANY of it. If I'm not going to do it immediately I put it on my TODO list.
> vim has like the worst default setting than any other app in existence
Laughing in urxvt
> Unless you have 5+ years of muscle memory (learnt with pain), all the shortcuts are too confusing.
Don't know which level of mastery of the default shortcuts you're talking about but speaking from my experience, going through vimtutor and oni[0] integrated tutorial everyday made me productive in less than one week, without modifying any .vimrc, granted I used neovim, but still...
SCM Breeze[1] is one of the first things I install whenever I set up a new dev machine. Setting up bash aliases for git commands like `ga=git add` is nice, but SCM Breeze takes it one step further and automatically numbers each file in the git output. Then you can do `ga 1 2 5`.
Asking lots of questions and thinking through the trade offs of more than one possible implementation before coding. Writing your implementation plan on paper especially helps find those unknown unknowns before you code. If you have trouble stating the problem and your solution in words, that indicates an unknown to investigate.
I.e. do a little writing to help avoid coding the wrong thing.
Learn how to use git's partial commit functionality `git add -p`, `git checkout -p`, `git reset -p`, etc.
Also get comfortable with rebasing. The most foolproof formula I've found is
git rebase -i <commit hash immediately before what you want to move> [--onto <branch you want to be the new base>]
Using the two of those together, you can clarify your commits and have code that is actually reviewable rather than stream of consciousness spaghetti commits.
Makes reviews more pleasant and looking back through history more useful.
Also learn to use `git log -p` to look through the change history of files. Immensely more useful than `git blame`.
i personally love partial staging/commits but haaate ’git add -p’. i use tig and a few other partial staging user interfaces. learning how to stage part of a line, in a favourite code tool, is a skill i appreciate for myself
Honestly these posts seem like the equivalent of "lifehacks" for devs to me :D
The only protip I have is to think through a task on paper before you start writing code. Scope it if unclear, sequence it into small chunks, do more technical design if its a complex change etc. See if you can reuse existing code/libraries to do what you need.
> 3. Learn how to use emacs in a tight code and test loop.
Came here to say this. Having a build delay for an average incremental change (edit a single file / compilation unit) be less than 2 seconds is extremely important for my productivity. Whenever I happen to do mobile development (this goes for both Xcode and Android Studio), having to wait 5+ seconds for the app to launch makes me lose focus.
You know those tiny useful functions to "parse any date into mysql friendly format", "check if string is a email or url", etc.
I have now hundreds of such functions in two huge files for the languages I work in most and it saves me a ton of time not having to reinvent the wheel or searching stackoverflow everytime.
Learning to write (simple) regular expressions will allow you to find and replace in text files fluently (capture groups are especially useful). It's a lot simpler than it looks initially.
I see many people suggest tmux, but I have not seen suggestions for my favorite terminal emulator.
Ctrl+Shift+E for vErtical split
Ctrl+Shift+O for horizontal
This is a killer feature. It is also much better than tmux because scrolling up is naturally supported.
I always keep an instance of Teminator running full screen on one monitor and whenever I need another terminal, I just split the screen, or go to the relevant already open post of the screen.
Having a visual overview of all instances of very helpful and much better than using tabs.
I think Terminator gave me a real boost on productivity. The split feature is great. And it has a lot more good things, like increase/decrease/reset text size, resize/rotate/expand or even drag-and-drop splits, grouping (to write in multiple splits at once), and the scrolling is done very well: scroll up a bit to stop following the output, scroll to bottom to follow again.
There is one single developer efficiency tip I can recommend: always dive deep.
While sometimes it makes sense to ask around for help on something, you should always spend 20 minutes checking the source of your favorite framework/company's library, and try to debug it yourself.
I developed my own gmail app script to accommodate my personal workflow, where all tasks and task-related information, including Gmail inbox, are organized and scheduled in the same hierarchy.
Here is why I think it makes me extremely efficient
1. I can group and nest together Emails, to-do’s, files and links that are related to the same goal, my lists are now much shorter and manageable and I can focus on my goals instead switching between gmail, todo’s etc..
2. Since my gmail Inbox is fully integrated with my to-do list, I don't need to manually and tediously maintain and synchronize them.
3. Scheduling a goal with sub-emails and tasks hide them and bring them back when I need them.This further reduces my clutter and makes sure I will not forget anything.
4. I can organize my plans and thoughts as I organize and drag inbox Emails, To-Do’s and Ideas into their related tasks.
5. Since every email has a natural place in my task hierarchy, I can quickly empty my Gmail Inbox and deal with email when dealing with their related tasks.
What made it stick for me was processing my gmail inbox (which I must do), inside my todo list, which quickly created a habit.
BTW I tried to market the solution and gave up, either most people think different or I am just a bad marketer.
There's many tools I've picked up over the years (sometimes on as needed basis, sometimes just for fun). The latter has been way more rewarding though in retrospect :)
Instead of enumerating all of them, I think this course page does a great job of collating them by topic: https://missing.csail.mit.edu/
I've been developing for 30+ years and I honestly don't see much of a productivity benefit using multiple monitors. I work differently when I have them, but not necessarily more efficiently. I find the head moving required to use multiple monitors more annoying than just alt-tabbing between windows. The only major exception is if I'm doing intense UI or visualization work and I want to see my code and the output at the same time. But even then, most of the time I can just resize my windows on a single screen.
Different strokes for different folks and all that, but, anecdotally, having been tech lead for various sized groups as well, I never saw a performance boost when people got the multiple monitors they always wanted, either.
I think part of it is that I find building a mental space of what I'm working on is the most important thing, and having tons of screen real estate acts as a replacement for that and thus goes against it, ultimately causing me to perform worse.
Subscribe to every slack support channel and mailing-list that your end user (or people dealing with them) report issues to. Getting not just the filtered bug reports but also the questions and not-really-a-bug reports will really help you hone into what your product is supposed to do, not just what you think it should do.
My tip is to change the requirements. Requirements are often poorly written and often contain a suggested implementation. Try to find the real problem, talk to your stakeholders and alter the requirements to find the easiest implementation.
I've saved months of work several times by doing this.
Using a modified international English keyboard[1] for programming.
My native language is Hungarian. We have 9 extra letters on the keyboard. Occupying the keys on the right side on the keyboard that used for brackets, semicolons, etc. on the US keyboard. These characters are moved to weird positions[1].
I use the English keyboard most of the time but modified it, so when I hit Caps Lock I can type Hungarian letters.
TextExpander. You can take your most common code snippets, table names, etc., and have shortcuts that expand to the full length. I used shortcut expansion around 10-20 times every day, which saves me about a couple hours per month (based on their internal analytics report).
..or just do not use variable names smaller than 3 characters if they do not mean anything, period. There is no difference in coding time and no good reason whatsoever.
I generally agree, but I think ii, jj, kk are fine for those iterator ints in loops, or the throwaway var in a list comprehension in Python.
The only other situation single-letter variable names make any sense is if you have a short function that implements a mathematical formula, like distance or something.
If you are working on the terminal a lot -- create handy aliases for common activities. For example, I have one to look at recently launched k8s pods. I have a "dracarys" alias which cleans out my environment and sets up a fresh test environment, etc.
That's what I've adopted a tiling window manager (i3wm) for.
I always use the same workspaces for the same things and have them open automatically on those workspaces, so most of the time, I just need one button-press to get to the terminal-instance I want.
For any API, truely understand the architecture and problems it tries to solve, and only then will you stop fighting the API and become productive with it. Assumption - the person who designed the API actually knew what they were doing.
Learning statistical process control. I now make control charts for virtually anything, both professionally and personally. It's an insane productivity booster in that it tells me what the meaningful things to focus on are.
Anywhere you want to find something, but particularly find something and do something with or to it, aliases or small scripts or plugins using fzf are great.
No specifics because I think it's a great Unix-spirited 'one thing well' and it's better to implement what works for you around it yourself, but I use it in: vim (filenames, file by contents, buffer names), password management, several git aliases for running commands with either a fuzzy found SHA or file, and more.
I was going to leave it at fzf, but as a bonus - bat. Use it in fzf previews, to see, with syntax highlighting, what you're selecting before you select it.
I play the Eleven Table Tennis VR game for ~1 hour at 9PM, become sweaty, take a shower and then I feel like I can very productively work until 5AM. Quarantine weird schedule, but it works :)
Often you want to fix something quickly but run into a stupid problem -- you don't have the right data locally, or the wrong version of something, or an unrelated thing is broken, or something is undocumented and the person who knows isn't immediately available.
Thus you feel deflated, the energy is gone, and finally you move to something else.
Instead, fix that stupid thing, once and for all. Get to the bottom of it, automate all the future times it would happen again, don't just hit it but hit it out of the park.
Not always, because you won't do anything else, but often.
If you are using vim, you should definitely try using cvim or similar for chrome. You’ll most probably stop using mouse when in browser.
If it’s about tools, I would suggest trying out i3 and emacs. I’ve used vim+tmux for a couple of years. Slowly I’ve migrated to Emacs. Somehow, I like it a little more than vim. I still heavily use tmux.
For tmux, having a startup script with pre configured windows and panes can be helpful as well.
Having a dual monitor setup can also be very productive.
I personally try to avoid using mouse. Most of my hacks are targeted to achieve this.
> There are a couple of annoying things with vimium which, I think, are general limitations for Firefox extensions, or browser addons in general: - it loads after the page, so if you mistakenly open a link, you have to wait until the page is fully loaded to close it with the extension, but you can always use the browser's integrated shortcut Ctrl+W. So far so good - Letters positions are not as accurate as let's say, qutebrowser, and sometimes when you open a link it's not the right one, even if the letter was indicating it. For a clear illustration of what I'm trying to say, try to browse stackoverflow and use the reviews queue. - When the page is saturated, letters start overlapping, you find yourself with one link hovered by 'A' and another 'AA' maybe a couple of links labelled: 'AB', 'AC', ... and in this situation it's impossible to select the link 'A', you have to do it with the mouse.
Thinking before I code. I developed RSI, typing hurts after a few hours. Now, before I code anything, I really think about what I want to type, so I only type the minimal amount needed.
Share more. Whilst it might seem appealing to be "the one person" who can do something, it stops you moving on to new things. I've seen so many people go from "the guru" to "the bottleneck" because they've never done this. Sometimes due to "holding onto power" sometimes lack of time or motivations. Docs and explanations help the author to move onto new projects and new developers to pickup the old projects - a win/win.
I've started using a debug diary, and it's been quite helpful. When I hit a problem, I often realize, "Wait, I've solved something like this before." I look through my debug diary and find entries related to what I'm working on now. It's saved me a bunch of time! Much better to read the solution I already came up with than to reinvent the same wheel for the 6th time.
Invest in your bash shortcuts! I have shortcuts for almost everything. To give you an example, I work in Rails + RSpec, each class has its own rspec shortcut:
`rspecs alexa_api` translates to `rspec spec/services/alexa_api_service_spec.rb`
`rspecm startup` translates to `rspec spec/models/startup_spec.rb`
I also have a bunch of other shortcuts for repeating tasks like expiring Redis, dumping a test database for local testing, etc.
I guess it's the fact that I first started with Norton Commander on DOS over 25 years ago and when Windows came around Total Commander was the closest I stumbled upon and I simply got used to it. Also Ghisler is still actively supporting it, which makes it very desirable for me. Otherwise I would've invent it myself.
[Edit] - reading a bit about Double Commander I guess it's what I am actually using on Linux. Both are written in Pascal and they look the same (well, as much as they can due to inherent OS differences). Didn't paid attention when I "apt"-it to its actual name :).
I sometimes like to leave my source code in a dirty (uncommitted) format where I've left a comment with what I intend to do next. I quickly jot this down at the end of a workday, or if I need to run to a meeting or have a short conversation with a colleague. This reduces friction for me to get started coding next time I'm back by the computer.
Hmm, sounds like you might be automating tests on the target, scripted from the PC. There's a place for this, but I'd call that automated integration/system testing, not unit testing.
Unit testing was traditionally considered impractical for embedded systems, because the code is so closely tied to the hardware. However, with proper architecture and unit test frameworks that simplify your life, it's quite feasible, and indeed enjoyable.
You write unit tests that exercise your embedded code, compiled for the PC, not the target. This ensures your code is portable.
I use the Unity/CMock/Ceedling framework when I'm writing in C, and CppUTest when I'm writing in C++.
I read and enjoyed James Grenning's book, Test Driven Development for Embedded C, but it didn't really click for me until I took the video courses mentioned above. In the second course, you code an entire simple but realistic embedded project in TDD fashion.
use AutoHotkey+SharpKeys to map macros, especially to keys on the home row. vim macros for recording on the fly. in general prefer keyboard shortcuts and command line tools.
regex is game-changing and regex101 is a great tool for it.
take notes and save snippets so you can refer back to them.
most underrated stack for productivity is proper diet + exercise + sleep.
Here is a real nice trick I have learn.
How to give Error Code uniqueness numbers to Errors.
That helps to find it later (find all) inside the code repository, know who code it and when.
The biggest efficiency tip I've found it blocking all distracting websites via /etc/hosts when my computer wakes. Website blockers never worked perfectly for me because of edge cases. Ex: YouTube is distracting, but sometimes I want to watch a YT video, so I would disable the blocker and then forget to enable it again.
The /etc/hosts based approach with a wake script to re-add all distracting websites has worked amazingly well:
There's this extension called Intention which, I think, handles these edges cases pretty well.
It doesn't really block websites, it just pops up in the middle of the screen warning you about how much time you've spent on that particular distracting page today. But you can mark a checkbox telling it that you will be using the site productively (watching a lecture at YouTube, for instance) and assign a period of time for it.
I've tried a lot of websites blockers that didn't stick but I'm really digging the approach of this one.
Make use of .bashrc (I'm sure other shells have similar).
For projects I'll be working on for a while, rather than CD into a nested directory, activate a virtual envirnoment etc. I have simple bash shortcuts that perform all the steps. Also for SSHing and many other common task.
On this note, (probably doesn't apply to many here) but if you aren't lighting fast with a terminal and are dragging files around etc. you are usually doing work in a sub-optimal way. Learning bash (or some shell, I don't consider I'm an expert) a little more all the time really pays off. Likewise VIM.
This is a cool ask HN. Here are some things I do for frontend programming.
Things I like to use:
- org-mode for structured note taking (debug sessions, wip tasks, architecture, rubber ducking, other team details, scratch buffer is like a giant clipboard manager, used to use notes for tracking time on tasks and starting + ending the day too but now only do if there's a crunch)
- isolated testing envs where I can test apps
- automated e2e tests via Sauce Labs + WebDriver with emails for catching regressions
- adding APM tools like email alerts when prod goes funky
- quokka + wallaby.js + personal development setup for interactive programming (yes proprietary but super helpful)
- personal scripts for automating work (little stuff like: run triage steps while debugging, control tools that have bad UIs, add git hooks so I don't do as much dumb stuff, etc.)
- using code formatting like prettier, stylelint, eslint (other teams contribute code to us so breaking their work locally saves us time in the PR later)
- pay attention to where you get stuck, optimize
- pay attention to what you burn large amounts of time on, optimize
- lots of little debugging things like look for adjacent problems after finding a new issue (9 debugging rules book is a good starter)
- thinking about Rich Hickey talks (lol)
- get unstructured time everyday to just hang out with other developers on your team and talk about issues or new tools to try out (favorite part of my work). Other people are really smart and come up with good ideas!
Things I want to get:
- source maps for all frontend scripts pulled in only for my browser, works for all envs at work
- full source code search across the company at work (large group, thousands of projects)
- automated a11y and performance tests to catch regressions early
- automating small code changes across hundreds of projects
- somehow get rid of webpack's complexity (pipe dream, yeah right)
Don't over-optimize. I work for a large tech company and I see many developers make the mistake to over design/architect solutions prematurely, setup meetings with colleagues to discuss how the solution could be leveraged by others etc.. But 80% of the work that we do is pretty straightforward and simple Google search will give you the answers. So just timebox few hours and come up with a hacky solution(proof of concept) that works. More often than not this solution is pretty close to what you want and you've not wasted companies time.
Using https://github.com/github/hub has allowed me to fire off small pull requests at a high rate. It _almost_ makes pull request creation as lightweight as commits. This leads to smaller pull requests that are easier to review and likely easier to get a quick review on for because of that. Additionally, the reviewer is IMO likelier to approve the PR, leading to me being a much more productive engineer shipping things at a higher rate.
There are a lot of advice out there and looking for advice that you dont need and mental models that you dont need can be a huge waste of time. Know the problem you have at hand, and find out what people are doing to solve it, dont look for answers for problems you don't have.
Reading non-fictional books can be a waste of time. IE: I've read over 40 non-fiction books in 365 because I enjoy them. This approach is not time effective because you can usually put the wisdom and knowledge of one book in 1 paragraph
Know your context and know if those advice apply to you
Using virtual desktops regardless of OS. Also assign hotkeys to them. I always find other developers wasting so much time with their mouse to focus a window when they could have placed a group of related windows on a designated virtual desktop which they can navigate to quickly with keyboard shortcuts. Virtuawin is what I use on Windows, but for Macs or Linux, you can configure the native virtual desktop equivalent well enough to make usable with hotkey. It cuts down visual clutter and gives you focus on a specific task way better.
Build admin and debug endpoints/pages into your services. Software is allowed to facilitate its own operation and troubleshooting, and it's almost always easier when it does.
Align your windows near the middle ok your display.
Displays are always in wide shape and if you maximize windows size of any app, you'll be looking to the left quite a lot.
What I do is I just narrow every window, so the left edge starts from a little bit left of the center of the display, so if I'm reading a page on a web, my neck would mostly be moving around the center of my natural facing position.
You can put other small window to the blank space at the left or put some desktop files there to be always accessible.
- learning more then a few basic hot keys for zsh (things like ctrl-k, ctrl-r and beyond)
- mastering tiling window manager, today I can switch, compose, mix and match windows in a blink of any eye - it used to take seconds
- switch to minimal OS setup, tile WM, your basic terminal tool suit, browser + the few apps you actually need (saves you trouble configuring all of that extra crap + updating it daily)
- find time to read MAN on every program you use daily, you will find lots of hidden gems in there
Use Makefile, Readme and .env files more effectively (mainly to avoid WTF moments when you don’t remember why you did something when you did it).
I am not good at remembering commands particularly when you have to deal 10 different technologies (Kubernetes, Docker, Framework specific stuff) so create some standard wrapper functions as make shortcuts and document them in Readme.
Just in time compiling means, you can modifiy your programm at runtime. I had wasted so much time in the implement-compile-run-debug-repeat cicle. Just the naviagtion to the part i want to debug had me cost a lot of time. With JIT you can implement and debug at the same time. Debugging every single change and seeing the effects helped me to avoid a lot of bugs.
I recently learned iTerm2 (on MacOSX) allows me to select and copy text from previous command, using keyboard only. https://kevinjalbert.com/iterm2-mouseless-copy/ I've always been annoyed having to reach for a mouse for this.
Started my personal wiki using dokuwiki. I sync it between my Ubuntu laptop and Android phone using Syncthing (on the phone I use Nginx on Termux to serve the pages). Very satisfying to have my notes available everywhere without having to share them with cloud etc (I already backup everything locally and on Backblaze so they are safe anyway).
I use a 4k TV as a monitor - I can easily see full size my code editor, terminal output, a page of documentation and the web app/site I'm working on all at once. It cuts down tremendously on the back and forth switching and reloading of things that don't feel like they take up much time but do materially eat into things.
Or better, use 3 4K screens :-) My record of the number of source files being opened concurrently was 10 when I was trying to understand a particularly nasty spaghetti code base.
Would you mind sharing the model? I tried using my Hisense but it wouldn't work. And I've read that TVs won't always work properly as monitors, but then obviously there's a much wider and cheaper choice of TV panels than dedicated monitors.
You are aiming for something that supports both 4K (3840x2160) [or greater I suppose] and a 60Hz refresh rate in that mode. On cheaper models this can be found on a HDMI2.0 input. More expensive ones might have DisplayPort as an option.
Shop around during TV season and you can find a quality model for $200/ea.
Like tmux in terminal, i3 for every GUI app. On a dedicated screen / workspace I run apps like browser, slack etc - and having them split the space automatically is super convenient.
Instead of commenting your code once in a while make it default.
I started always commenting and documenting my code (even the simplest parts for consistency) and it dramatically improved my productivity when I revisited if afterwards.
I used to use AutoHotKey several years ago. It provides remapping hotkeys, text expansion (snippets), macro on operating system level, etc. But it only runs on Windows. :( It's the very few tools on Windows I miss.
Now up and down arrows will search your history with whatever you already have typed at the prompt. Re-running a command often? Probably just have to type its first letter and hit up arrow once or twice.
Spacebar (pagedown) |
Shift + spacebar (pageup) |
Ctrl + tab (cycle between last recent opened tabs) |
Cmd + number (go to tab of number) |
Cmd + K (go to search box) |
Cmd + L (go to address bar)
BetterSnapTool for snapping windows on your computer, been using it for years. I'm assuming every dev takes advantage of screen real estate, but if you aren't, you should.
For me, having OhMyZsh installed with a couple of plugins such as: git, docker, vscode has greatly improved my efficiency as the amount of typing has reduced considerably.
The Pragmatic Programmer, which was recently published its 20th Anniversary Edition, talk about these tips in whole book, from The Basic Tools to Pragmatic Starter Kit.
(Really, use any fully extensible editor and take the time to learn it well. I'm told Vim fits the bill too; my personal experience just happens to be with Emacs.)
This is one of the things I love about macOS -- when it works. I really wish they worked better.
Yank works 90% the same, which is close enough to be useful but different enough to be annoying, daily. Repeat count is disabled by default. Transpose is broken in Xcode 10, and I heard Xcode 11 fixed it but one of my Macs can't run (the macOS required for) Xcode 11 so I'm hesitant to upgrade.
And of course if you use Firefox, they've implemented the whole GUI from scratch, and didn't bother with half of those, despite bug reports having been filed literally decades ago.
I get the feeling that Emacs keybindings on the Mac are like map/reduce in Python: it got snuck in version 1 by a fan, and the maintainers since then haven't really understood why it's there or what to do with it. I fully expect it to simply disappear one of these days.
This is really interesting. What key bindings do you use consistently? I wish the vim worked the same way. I'd love to use vim nav and modes for any text in any app.
1. I use Visual Code with carefully chosen extensions that alert me all the time about error.
2. I have enabled linter autofix on safe. It fixes a lots of stuff very good fast.
3. I wrote several own extensions. One for example shows error in the code when i do some mistakes i normally do. It also boosts my speed because i dont go through the whole process of finding these errors.
4. Since i code using TypeScript and Angular i use a lots of OOP and composition to reuse easily and modify code fast. And have less errors. OOP gives great propert and method suggestions in VSC so there is less space for errors.
5. I have a lots of generators and snippets custom for my projects. So writing similar things is blazing fast.
6. I dont overcomplicate stuff. Write everything as simple as possible so it is easy to refactor and modify and understand.
7. I use bash with a menu to have complicated operations like git operations to be done easily. Without clicking with mouse. It is a boost.
8. I have written also some software in C# that operates on the TS code base. I use it to generate some more complicated stuff that includes multiple files at once. I use code comments to put for example snippet examples into Angular code. Than a C# app looks for these in whole project and generates snippets for VSC automatically.
9. An extension for VSC executes tests one by one. I can choose test to start. It is faster than start all each time.
10. CI/CD, unit testing, linting, end to end tests give a boost if done right.
11. I side from solutions that are hype but cause loads of code to be written. For example some silly state management libraries. Overkill.
12. My PC is packed with best hardware, updated regularly to be fastest as possible. Since every second you wait ia a second you waste. PC ia better because you can upgrade it cheaper than laptop. Remote access is the key if you want use laptop.
13. Good sleep and diet, sports is essential for good performance. And music.
14. I never go into the zone if i am not 100% sure what and how to do stuff. In the zone you arw blazing fast, but be careful to not write useless code because of bad approach.
15. Notes notes notes. For everything. Refactored often. JIRA as a single source of tasks. Notepad as a detail task list.
16. I adjust tasks to the time of day and my mental focus. Some easy task for first half hour, and than the big ones for the rest of the morning.
17. I have wrote a lots on keyboard to have very high speed. It saves me time writing (actually you need to use keyboard when you are a dev anyways) and less wasted time for typos.
18. All notifications off. All spam sms, emails, calls off. No distractors.
19. I never code when i am not 100% sure what and how to do it. No guessing, always asking. A lots of wasted time saved. Never being ashamed to ask even the most stupid question.
20. I always remember that things take time and not stress too much about it. Stress harms performance if too high. But also some stress can let you go into the zone. And this is magical place to be in.
Bonus: when you have the highest effeciency, the next step is to help other boost theirs. I think this is something important too. Some people get to the 10x point and think itis enough. It is not. Your team has to be too. If you are the only one, you did only tiny part of your job
Over the past ~half year I gradually established a workflow consisting of a simple directory structure, a handful of bash-aliases and chromium profiles. It looks like this:
Each project gets its own directory somewhere in my $home with some subdirs. There might be various others specific to the project-domain, but these are the most common:
• notes/ contains plaintext files of everything i write down. File-name structure is <date-str>:<note-type>, e.g. 2020-05-09:call (:call are the most typical ones). As others have also suggested, I take a lot of notes for record keeping & plannning purposes.
• workspace/ contains all source code repositories, or other files necessary to work. Pretty much anything I might touch while coding.
• auth/ for things like ssh public keys used on the project
• downloads/ Project-specific downloads.
Then I create new entries in my ~/.bash_aliases that "get me" right into the coding-env of a project. Say I work on a python project, I might have an alias that looks sth like:
alias fb='cd ~/fooproject/workspace/foobot && echo "loading virtualenv..." && source .venv/bin/activate';
On of my current side-projects is a full-stack clojure(-script) app with these aliases:
alias foh='cd ~/opendecks/workspace/foh';
alias fohj='cd ~/opendecks/workspace/foh/src/clj/opendecks/foh';
alias fohs='cd ~/opendecks/workspace/foh/src/cljs/opendecks/foh';
First one gets me into the the source-root of the project, the other two into respective sub-paths for back- and front-end. These aliases allow me to immedeiatly "jump into a project" from any open terminal.
Lastly, I started employing chrome/chromium's user-profiles to seperate browsing for each project. This isolates work-related browsing from my personal activity (cookies, logins, passwords etc), and provides a nice cognitive barrier against procrastination. I'm not logged into any social networks on those work-profiles and switching to my private profile to browse facebook is a more concious act than opening a new tab in the current window. Each browser profile is set-up to save all downloads to the project-specific `./downloads` directory, again not to mingle (potentially confidential) resources.
I haven't automated any of the above steps, because it's so simple and takes me at most 10-15 mins to create a new instance. This has worked really well for me so far, because it clearly seperates almost all resources associated to a project, while being friction-less in tooling/switching between workspaces (versus e.g. waiting for a VM or docker instance to start). Another nice side-effect is that I can easily archive or clean-up past projects.
Learn to touch-type properly. I'm not a shit hot typist or anything, but it does make hours on a keyboard more pleasant. And (hate me) I do judge other devs on it. If someone can't be bothered to invest a few tens of hours on such a key productivity enhancement then what does that say?
I think this is often undervalued. It's not necessarily the typing speed, but the far larger slowdown caused by distractions like error correction, using the mouse, getting irritated and taking a break etc. Meauring words per minute in typing static text doesn't really show that up.
Also, it's something that's hard to appreciate the value of until you experience just how big a different it can make.
Strong typing skills also form a good foundation for learning Vim and other editor/stystem wide shortcuts, which can make the whole experience of interfacing with the computer an order of magnitude more efficient and enjoyable.
Fewer distractions means more uninterrupted thinkning time and better code.
I am unimpressed with this site. It seems to use a large number of made up words (risittle, ...). The words it chooses can even be one letter off of a common word (itsell -> itself).
It literally says that it uses random words, and explains why, in the website description:
> Instead, Keybr.com generates random, but readable and pronounceable words using the phonetic rules of your native language. These words look almost natural, and often they really are. Typing sensible text is much easier than repeating random letters, and it helps you to remember frequent key combinations. The latest point is very important.
Since public schools closed, I've had my 3 kids (11, 8, and 8) practice typing 30 minutes every day. Easily one of my best parenting decisions ever. :)
Same. Learn to type ~100 words/minute, I was surprised by how much more productive it has made me. Using a touch type trainer for a minute or two can also be a little mental break from writing software for hours.
If you don't already have one get a good ergonomic keyboard. I recently switched to an ergodox ortholinear layout and it immediately exposed my bad typing habits. The split hand layout saves my neck and shoulders and my working posture has noticeably improved. Took about a month to get back to 80 wpm and don't feel like I've plateaued yet trying to get to 125 wpm.
I had a Judo coach who used to scream "Practice doesn't make perfect. Perfect practice makes perfect." In practice sessions focus on not making any mistakes & using only the correct fingers to strike keys. It can be hard to resist the temptation to start hammering as fast as you can in practice but slowing down once in a while can greatly improve muscle memory.
I use https://www.keybr.com/ to track stats, practice limited or different key sets, it also supports non qwerty layouts for stretching your comfort zone. For touch typing practice it displays a real-time keylogger map so you can auto-correct your finger positioning without looking at the keyboard, if you've got a fancy keyboard (Shout out Arts and Crafts HN) you can hook up the firmware to do the same.
There is a difference I think between typing speed when you are transcribing vs writing your own text.
I find that I get pretty slow speeds when doing typing tests, but most of the time it's because I need to stop every word to figure out what the heck it is wanting me to type.
When typing out my own stuff, I don't have that break because I already know what I want to say.
100 wpm when no-look-pecking, 60 wpm when no-look touch typing.
Feels weird man. xD
Also, I have a friend who used to be a court typist. Does zero coding. Borders on comically-computer-illiterate. Types like 170 wpm. So jealous. Such a gift, so wasted.
Just as a shameless plug for anyone reading the replies - I'm working on creating a touch typing course for developers here: https://typeright.herokuapp.com. Still lots of work to do (hence the lack of domain name) but I hope to finish it in the next month when I have time. https://typingclub.com is great if you just want to learn to touch-type in natural language but I found it and other courses didn't help my typing skills for coding that much so I'm building one myself.
What's the value add of touch typing? I tried it for a bit, but it seemed like a waste. I can consistently do ~90-100 WPM on https://www.livechat.com/typing-speed-test/#/ with very few mistakes just typing with 3 fingers on my left hand and 2 on my right. Feels like input speed is one of the least important parts of coding.
Sounds like you are touch-typing, you just don't use the home-row standard. Neither do I. There's plenty of other ways to do it, they're just not taught classroom-style so people like to discount them.
The benefit of touch-typing is being able to get your thoughts into the computer without interrupting said thoughts to figure out where stuff is on the keyboard. Keeps the flow going.
I was hoping your image would be a video of you showing that you're actually two coffee-crazed squirrels.
You must know that average touch typing speed is 40 wpm, professional is 75 wpm, therefore you have nothing to gain by optimizing further. Especially since bragging about doing it the hard way lets you seem like you'd be even faster if you tried.
For a start you don’t have to look at the keyboard so you can spot typos as they happen and take advantage intellisense etc. Although at 90 wpm you’re probably not looking at the keyboard either.
For me the main reason I value touch typing so highly is due to the reduction in cognitive load. I think things and they appear on the screen without any conscious thought. I honestly couldn’t tell you what fingers I’d just used to type the words.
Good technique helps prevent injury. Rheumatoid arthritis and carpal tunnel are no fun. Adding “it’s going to hurt” to the list of reasons not to sit down and knock out that feature really isn’t helpful to your productivity.
I ended up tracing my pain to lazy technique in my right hand. Sorted that out and I’ve been pain free for years. It was _no fun _ before I figured it out.
The biggest benefit in my view is not typing speed but keeping your eyes on what you're typing instead of on the keyboard. You catch more mistakes that way. If you're already keeping your eyes on the screen, you've got most of the value already.
A lot of people type without having to look buy without using the proper technique, essentially doing hunt and peck with two or three fingers but with good muscle memory. Proper touch typing is about using specific fingers to specific keys and remaining on the home row when not in use.
Yes, that was me! Two or three fingers per hand. I think a lot of us do like that. I learned the proper way like a year ago. I sill had to force myself to put my fingers on the home row. The problem is I use their laptops in weird places like on a couch or bed where proper touch typing is just not comfortable and it easier to type one handed.
Steve Yegge's blog[0] on the matter made me convinced to learn to touch type. Went from ~60wpm to ~90wpm over a few years of practicing every once in a while.
I've used Dvorak for the past few decades. As suggested by one of the other comments, it took about 2 weeks to learn well enough to get back to my normal speed (>80wpm).
It makes a massive difference in comfort. 5 minutes typing at full speed on QWERTY and my wrists start to hurt. (I really don't know how people do it.) On Dvorak, I can type all day at full speed with no pain at all. Surprisingly, the keyboard hardware itself doesn't seem to matter.
(Details: I mostly use a Mac, with caps lock mapped to control, NSRepeatCountBinding set to ^u, and -- obviously? -- I rarely touch the arrow keys.)
> 5 minutes typing at full speed on QWERTY and my wrists start to hurt. (I really don't know how people do it.)
Wrist-bending to maintain the home row, I suspect. I have a different, self-taught style that partially involves holding my hands above the keyboard and using muscles all along my arms to move my hands without bending my wrists much - never had any problems.
Moved to Colemak and learnt touch typing on it directly (to take advantage of the mental break). Really improved my "writing feeling", solved some nagging RSI issues and in general helped in "well-being". I don't type particularly fast, but I type way smoother, which is what feels good. I can't recommend Colemak enough, although I have never convinced anyone (in around 3 years).
Greetings to a fellow Colemak user. I've used Colemak for about 7~8 years now. It's really interesting to see home row keys polished to a shine, while others aren't.
I do always change one thing, and recommend it for others too, is to change caps-lock into a control.
For a long while I've had C-M-A mapped in caps lock, then I moved to using a Gergo (60% split ortho) programmed with QMK, so it's now mostly my own layout, specially for modifiers (see [here](https://mostlymaths.net/2020/01/new-keyboard-gergo.html/))
Yeah, it definitely did, re-learned to touch type in Colemak and also upgraded to ortholinear split keyboard. Yet it's takes quite some time after ~20 years of touch typing on Qwerty.
I use Dvorak exclusively and echo the other comment -- I'm not sure that it has made a dramatic difference in my speed, but it is much more comfortable! Your hands sit neatly on the home row, and do not have to reach much.
The only downside is that I use vim and after many years of Nethack am pretty good at hjkl movement, and I can't even take advantage! :( Definitely something to consider though if you want to switch and are a vim user.
I use QWERTY but on an ergodox. Massively improved my experience at my usual workstation and it’s still seamless to jump on a normal keyboard if I’m using my laptop or someone else’s machine.
Node.js: you can share code between client and server side.
Typescript: you don't need to remember the arguments of a function or return types. Easy to refactor. Just no brainier. I also use it for all my config like webpack config with ts-node or ts-node-dev for auto reload.
Auto format all your code with 1 command. Prettier or lint fix. But that's basic.
Don't ever write global css. Use emotionjs or styled components. Scoped css it's a delight.
If you notice all my tips helps you to reuse the code you write. Therefore write less code, more efficient.
Plan all your stories ahead. Then do one at the time. Then mark it done. Prioritize.
Avoid being an early adopter of new technology despite everyone on HN raving about it. Instead focus on solving actual business needs that grow revenue or profit margins.
This might be a little underrated by the average developer just how important this is.
That isn't to say, write code fast, even if its garbage. Good clean code is still a virtue. I feel like sometimes people conflate the two. So to add on, don't do that!
1. It helps me stay focus on the task at hand. One recording for each task 2. It lets me practice how to articulate stuffs. Its like blogging but ephemeral (because i wont upload this) 3. It helps me get motivated. Cant let my "thousands" of viewers down 4. It lets me review my blunders so i know how to watch out for them in the future 5. Its fun. I can let off some steam because i can curse all i want, my audience is 18+ and fine with some "sentence enhancers". I am Filipino but i record in English so i get that extra bonus practice.