Ah yes, working on multiple files was an issue until very recently.
Firstly, at claude.ai you can upload multiple files, so Claude will take those into account and even suggest changes to multiple files. You are then, however, still copy/pasting from a web interface.
Enter cursor (https://www.cursor.com/), you can either use a Claude API key (but it will warn you that all the features they want you to pay for then don't work), or just use the free version, like I am currently. It gets me enough prompts per day to improve my life.
Or you could pay for it, but I have a feeling that this is a sort of WinRAR situation...
was pretty good too, but it had taken on too much in its latest version 2.0.0. In other words, it is too unstable at the moment (but probably worth looking at again later).
It's not even the same class. Copilot does autocomplete essentially. Cursor does that but with the scope of a file/project. That means after you change one thing it will offer to skip to another part of the file and do the other part of the change there. Refractors often end up being a single change then (tab), (tab), (tab), ... And it's correct a shocking amount of time. The UI of proposed changes is also better.
I'd recommend just trying it, because it's hard to summarise how much it's not copilot.
The "tab to next change" feature is amazing and I would love it if it worked reliably.
As it is though, I suspect whatever model Cursor Tab is using under the hood has a fairly small context window, so the range of that "tab to move" feature ends up being pretty limited.
Overall my takeaway after months of using Cursor is that it has some really promising features but their LLM engine is too underpowered to deliver.
Hopefully that will change over the next few years. The potential is definitely there.
I've completely switched over to Cursor from Copilot. Main benefits:
1. You can configure which LLMs you want to use, whereas Copilot just supports OpenAI models. I just use Claude 3.5 for everything.
2. Chatting with the LLM can produce file edits that you can directly apply to your files. Cursor's experimental "Composer" UI lets you prompt to make changes to multiple files, and then you can apply all the changes with one click. This is way more powerful than just tab-complete or a chat interface. For example, I can prompt something like "Factor out the selected code into a new file" and it does everything properly.
3. Cursor lets you tune what's in LLM context much more precisely. You can @-mention specific files or folders, attach images, etc.
Note I have no affiliation whatsoever with Cursor, I've just really enjoyed using it. If you're interested, I wrote a blog post about my switch to Cursor here: https://www.vipshek.com/blog/cursor. My specific setup tips are at the bottom of that post.
Do they impose arbitrary time based restrictions (that wouldn't otherwise exist) when you use a Claude API key instead of paying? I went back to VS Code after something like that (seemed to have) happened.
Honestly, once you learn the copilot specific hot keys you can do all of what cursor does and more. in fact there were times that i felt the team at VS code clearly could have added features that Cursor has but chose not to because they led to more unwanted code slipping through.
I did like the edit tab completions from Cursor but not worth 20$/month and guaranteed enshittification
Hugo might have the worst documentation I have ever encountered in the wild.
Which, if you think about it, is the height of irony. Hugo, the thing so many people use to make documentation, has terrible, terrible documentation. It is impenetrable.
This is coming from a Go dev who understands Hugo's internals quite well and uses it in production.
Had I known what my journey with Hugo would be like, however, I would have never touched it. In all that time, with all the headaches it caused, I could have just built my own SSG.
The community is very good indeed. That was always where I ended up finding answers.
I don't know whether I have seen worse. I may have seen 'equally bad', however.
There's a good chance that I'm just not good at understanding that style of documentation. I feel there is a huge lack of examples. The examples given are very specific and they do this thing where, when showing two examples of the same functionality, they change all the variable names and data-structures, making it impossible to follow.
But that may just be me. Some people might find it useful to see the same example with different data-structures and variable names.
It’s pure bliss compared to just about any Python library. Oh, you want to know what arguments this function takes? Well, a kwargs dict, duh! Why not just read the full example section to hopefully find some of them?
Imagine my surprise when Hetzner Cloud recommended I use Ubuntu for my server (prior to the enterprise upgrade shenanigans) and then, suddenly, it tells me I don't get upgrades until I pay. On Linux.
Ubuntu has a lifetime ban from me. No thanks. Whenever I setup my server again (I can't face the prospect, as it'll take me weeks) it'll probably just be Debian.
Ah yes, we are back to fetishising over-work for no reason.
I sort of understand if a CEO wants to work 7 days a week. Go ahead. If it's worth hundreds of millions, sure.
Normal workers? You'd be nuts to let anyone do this to you. We should be working less, not more.
The arguments against it are all the same: You can't work anyone effectively for 7 days a week 10h+ per day, so it just becomes an exercise in control / loyalty / devotion to the higher ups (who, as you so rightly point out, don't actually put in anywhere near that much work).
Why governments aren't regulating this more, is beyond me. In Europe you'd probably struggle to do a 7 day, 10h+ work week legally, but in the US it appears to be fine.
Why YC isn't enforcing better standards is another question. I suppose, again, that it's extremely skewed to the founder's perspective and to them it's obviously worth it.
One more upvote for Dokku. Been using it for as long as I can remember hosting things on servers. It is such an incredible piece of software. And open source to boot. If any of my projects ever make money, Dokku will be the first project I'm funding.
Even though the two may not be fully comparable, I would like to give another vote to pocketbase, simply because I've found the experience of working with it to be absolutely stellar.
Firstly, at claude.ai you can upload multiple files, so Claude will take those into account and even suggest changes to multiple files. You are then, however, still copy/pasting from a web interface.
Enter cursor (https://www.cursor.com/), you can either use a Claude API key (but it will warn you that all the features they want you to pay for then don't work), or just use the free version, like I am currently. It gets me enough prompts per day to improve my life.
Or you could pay for it, but I have a feeling that this is a sort of WinRAR situation...
reply