Hacker News new | comments | show | ask | jobs | submit login
No software development process for startup - What do you think about this?
9 points by anonymous-dev 1853 days ago | hide | past | web | favorite | 20 comments
I have been pondering on this for a while and need to get some feedbacks from fellow developers.

Here is the scenario:

I am a developer documenting what I work on (apis, tools, etc) to. However, my colleagues don't. This have been driving me nuts but I cannot change them (I tried).

I also like to improve existing tools (build system/ code base). However, my colleagues don't.

Here is what every developer faces every day: - No automated build (e.g.: Continuous Integration Server) - Released archived as emails instead of FTP/Nexus like software. - Developer builds production releases on their dev machines. - Undocumented APIs/build system - No automated test - No clear roadmap - No code review - New developers are hired without going through any technical interviews nor have background or experience required (programming language, software dev lifecycle, mindset).

The business was not doing well until recently which explains why all those things creeped up (although IMHO, automated build and unit test should be done anyway).

Now, the business is doing well, I managed to develop an automated build system after a year or so battling with management and reworking tools to make it happen.

New hires are coming in but I feel those processes aren't really used/sought by other developers. Some management folks seems to care about the issue but don't see the need as much as I do; a strong, well structured yet flexible software dev process is a very valuable asset for any company and should be respected fully.

No other developers (about 20) raised any concerns about the current software dev process, only me.

a) Am I too impatient since things are slowly changing ? b) Am I too process oriented ? This is a 5+ year old startup. c) Am I the problem ? d) Is this software shop not a good place to work for and improve one's skills set ? e) Should I just suck it up ? That's the startup world.

I really appreciate your feedbacks.

Have you read the Pragmatic Programmer? They talk about how to handle a situation like this. From what I remember it said to build a pocket of excellence in your dev workspace first, and then expose everyone else to it.

Set up the CI/build server on your machine, have it email everyone when the build fails. They will see the use of the tool then, and when the team accepts it they will hopefully get an official server to host it on.

Write tests for your code, and when other devs work with you on issues, or break things, write tests and send to them "Hey, i wrote this test that shows the code you wrote is failing, can you make it pass?" Also ask others to make unit tests. You will have to write the bulk of tests. You probably know which others on the team are open to writing tests and working with you on the pocket of excellence. Get them involved first, and then the idea will spread. A good idea would be buy them copies of Pragmatic Programmer too.

No, you aren't process oriented. A software team that doesn't at least like the idea of automated builds, and tests is crazy. As far as if it's a good place to work and improve one's skills? That depends. In some teams you might never get them to adopt the pocket of excellence. If you are working in a place like that, then it's probably not a good place to work. But if you have a shot at converting them by just doing it on your own first, then it would be a great opportunity to get experience with setting up these aspects of a software shop. And you can put that on your resume, blog about it, etc. In that case it would be a great experience.

I heard about the Pragmatic Programmer book. I guess it is now time to read it.

About CI server, it is on its way but only for production release. No continuous build for now mainly because there is no unit test infrastructure setup.

About writing test, well, no one uses unit-test for the reason above. It maybe a good time to introduce that and get to show them the light :)

"In some teams you might never get them to adopt the pocket of excellence. ". I am a bit afraid of that. Why ? I have spent some time trying to change things and yet I see little or no effort from others.

What are you thoughts on documenting things ?

Very high level documentation is good. Things like, what are the steps to get the app to build, what tools do you need to install, frameworks, etc. What's the directory structure, so that a newcomer knows how to approach the project.

I'm a fan of self documenting code. Trying to document code is nice, but rarely possible to keep up to date. I do like to put short comments on a method if the name and class its in don't provide enough info. It's better to spend time writing code and tests than documentation.

You can't always convince people to do things they don't want to do. Some people don't care about putting more than the minimum of effort into work. Most people hate change. But if you just go ahead and create the pocket of excellence, people tend to notice why its good and if you get momentum then the rest will follow. Who are the influential people on your team who if they started using these good practices, the rest would follow? How do you think you should best convince them? For some people, find out what motivates them, and then see if you can explain to them how adopting these practices would benefit them.

About the first paragraph, that's exactly what I am trying to get going.

About the second paragraph, I 100% agree with this.

I did create this pocket of excellence (myself only) and lead the way for some time. Yet no one notices it and the ones who do, well, they don't care about it - both for junior and senior developers. IMO, the most influential peoples are so called "rockstar" developers within the company and they clearly stated they don't see the interests. I already talked to them and attempted to convince them. I am running out of ideas.

Why are the rockstar dev's not interested in things like automated builds, and unit tests? Are they really "rockstar" dev's? It might be political. Maybe they like being the only ones who know how to build the product, or that it has to be done on their machines.

Sometimes people shoot down ideas of those below them because they don't want to see a "threat" rise up to compete with them. Maybe if you can't convince the rockstars, just get everyone else onboard.

So you created the "pocket", are you getting anyone else involved? If you already tried to get official adoption of these practices and that didn't work, I dunno. You might just get it done. A lot of times managers ignore requests like this because they don't want to bother with change. THey don't realize the benefit, or they don't want to try to force change on a team (maybe they don't want to adopt anything the rockstars aren't into).

"Rockstar" developers are simply jack of all trade developers with the most experience by far in the company - they created the entire system from scratch. About it being a political thing, I believe so.

I failed to get everyone else on board (even new hires as per my last discussion) because they seems to have already pledged allegiance to the "rockstars" AND/OR don't give a fuck about good engineering/project management practices. Eg: Some have already told things like "I don't mind having a commit fixing two issues at once while the bug report clearly specifies to fix only one of them. I don't mind not raising an issue in the bug tracker about the second bug I silently fixed".

Some managers are former developers who did implement those practices while they were devs but they don't see the urge now. Maybe they know they will not have to fix the issues directly and that might be the reason why they don't push those changes as much. What do you think ?

Well, there is no guidebook for this kind of thing. I don't know what to tell you. THere are probably people who know how to accomplish what you are trying to do, but I don't know how to advise you past this point.

The takeaway is that while you sound like you are probably a solid developer who's passionate about good software dev practices, the thing you need to work on is people skills, and the politics on how to get what you want. Basically you have a set of things you want to see done from a tech perspective, but now you need to figure out how to effect that change in the company you're in. It's not as easy as it sounds. It's not as easy as explaining the benefits to people. A lot of times, to be honest, I've felt very passionate about something and when it came time to try to convince others, I couldn't really convince them and it was really frustrating. The only thing you can do in those cases is to learn to be less rigid about it. If it's going to bother so much that youa re going to keep stickling about it, you might better off working somewhere else.

I learned that I am sometimes very rigid in how I want things done, but that is only my idealized version of how thing should be. The rest of the team may not want to adopt my ideas, so I had to learn to accept that. There is the perfect way of doing things, an then there is how things are done in real life. Everyone has different motivations. Sometimes people just aren't into their software dev jobs, sometimes they just know software as they learned it years ago and don't want to learn to do things the "right" way, or don't want to learn new technologies. Sometimes the motivations are political. Sometimes it is possible. I think it's often impossible to effect change if the "rockstars" aren't in agreement, often as we discussed before, because they don't want the guys lower on the totem pole showing them up by improving things.

It sounds like you have reached a dead end. You can either figure out how to do it on your own, which will take soft skills. You won't do it by convincing anyone that your ideas are technically the right way to do things. People generally don't care about that in the real/business world. You will have to learn to influence people, to convince them to do things they aren't into. One of the ways I know of is to do the pocket of excellence and make the team dependent on your way of doing things, and then when they are dependent on it, expand, and formalize it. The CI server is running in a VM on your workstation. You just use it and set it up, and make sure everyone on the team can access it. When they want a build, you send them a link to the URL for your CruiseControl instance (o whatever CI tool). They want to fix a piece of code you are in charge of? If they code a bug, make a unit test and ask them to write code to make the test pass. Just set things up the way you want as much as possible an force people to use it. Don't give them a choice. If yuo try to have a discussion in which you envision people agreeing with your way of doing things, that will never work. Just force them to move in the right direction by doing it. Managers aren't as powerful as you might think. The team generally does what they want. The worst that could happen is they ask you to cease and desist, but they probably won't because they probably know that you are a good contributor and they will have to put up with your desire to do things "right".

That's it, basically is to keep trying, but don't try to argue verbally, just put stuff in place and don't give people a choice, or you can find other employment. Focus on small things, and 1-2 changes at a time, you don't want to change everything at once.

Hope this helps.

Personally I'd probably leave. I think there is a point where its healthy to try to change the system but it sounds like you are fighting a current that is too strong to encourage the way you want to work.

I believe you when you say your efforts are needed - the problem is management very frequently has a hard time being able to understand that there is an upfront cost that will eventually pay off. They are looking at spreadsheets and just want to see that profit column keep on going up - you essentially are proposing that the profit column should go down a bit (development costs) with the promise that it will eventually go up much faster.

I've found that software companies either qet the necessity of quality/process or don't (also some types of businesses do not really need much software quality). Most of the time this is determined by if they have technical people in the leadership ranks or not. By the hiring practices you listed Im guessing they don't get how software quality and process impacts the bottom line.

My pessimistic view is that very little will change with the exception of burning yourself out - sorry for the dire prediction Ive just been there myself.

"Most of the time this is determined by if they have technical people in the leadership ranks or not.", can you elaborate a bit on this ?

According to my experience, management has always viewed software as binary world: it works or it does not.

About your last sentence, I was about to but gave myself a reason no to; instead of trying to change things, I simply leave them the way they are and stop "bitching" about it. Either way, I will see if good engineering practices are really important or if a team can do without it (and what the potential cost would be).

Now, my only focus is becoming good at spotting bugs, fixing them and implementing features with fewest lines possible (a.k.a Spartan programming) :)

> "Most of the time this is determined by if they have technical people in the leadership ranks or not.", can you elaborate a bit on this ?

From my experience, non-technical managers often don't realize the importance of these processes. Us programmers know how important is to control how the codebase grows—when there are more than a few programmers touching the code, and when the software keeps growing, a good test suite will prevent the team from having to spend a lot of time trying to change something in the future without breaking anything else (often in vain.)

I worked for both kind of people, those who understand and those who don't. I found it was a huge PITA to work for people who don't understand, especially if they can influence the hiring process. They will have a direct impact on the culture of the team, and I don't believe a single developer can change this alone.

This is just anecdotal, but the place where I worked for this kind of people just closed the doors; the software grew to a point where it was unmaintanable and, despite all my efforts to convince them on fixing things instead of putting new features like mad, nobody ever listened. Now talk about frustration...

Edit: of course I don't believe every non-tech manager is like this. There are those who, despite knowing about nothing when it comes to software development, knows how to identify someone who does, hires this person and trusts their opinion. Basically, people who know what they don't know.

> instead of trying to change things, I simply leave them the way they are and stop "bitching" about it

Please don't do this to yourself. This is likely to go one of two ways:

1: You stop caring about quality.

2: You forget what quality looks like.

The end result is more or less the same: the quality of the software you produce is likely to go down.

I have had very similar experiences before and watched colleagues take the same path you're proposing. Whilst they may be happier in their positions, their QA goes right out of the window and things start getting even worse.

It's better to have a reputation as somebody who doesn't tolerate shit, than to have no reputation for anything other than shipping crappy software.

Don't worry about the process. Worry about the outcome of the process. Is the site always down? Do you fail to ship? If so then you need to fix things. Otherwise just let people work the way that lets them be most productive.

For instance, I'm not a fan of unit testing. I find that 99% of the time when my tests fail, the failure is the test itself, not he code. WHen I'm forced to write uni tests for everything, it slows me down greatly. I prefer to keep my code clean and readable, instead of focusing on 100% code coverage. The same principle applies to CI and other such "processes", in my experience. But thats just me, everyone works their own way.

I agree with this, if everything works with the company's current process, there will be no way to convince people to change, and there wouldn't be any reason to.

Try to maybe focus on fixing problems that your company actually has.

Will do that.

Are you a single dev or do you work in a team ?

I recently quit my job for pretty much the same reasons. (I even created an auto build system, too... weird.) I actually faced a professional development dilemma: No forward progress. In an environment like that, where you have 20+ other people who don't care, aren't paying attention, just hack things together... you're not going to make forward progress (professionally, and the business will eventually be burdened with wanted growth and a messy code base, or something will break big-time and cause major loss of revenue). They are and they will continue to be stuck in the muck until someone closer to the top most likely runs into a very hard decision, or looses a great deal of money and has to face up to the reality of how they're building their business. I recommend moving on. There's some positive to be said for being pragmatic (sometimes you just have to hack it). However, if this culture is persistent, it's a BAD culture, especially if they hire inexperienced / unqualified people, and start teaching them these habits / traits (bad ju-ju for everyone, the entire development world, when places operate like this).

"Now, the business is doing well"

That is what matters to the decision makers in any business at the end of the day. So if that is the case, you might not be able to convince the management. Sure, your points are valid and I have the same frustration at work many times but understand that management doesn't care about proper unit testing etc. unless it impacts the business. As someone else pointed out, if you have critical production bugs frequently causing loss (specifically monetary), then you have a solid case to argue to change the process.

Not to say that you should just suck it up. I think it is worth to try and do things the best way but just keep in mind that you should not expect management to just do it because it is the right way to do it. So keep expectations low but keep trying as well.

This reminds me of "Happiness = Expectations - results". You are spot on.

I think I mixed up my personal expectations (as a solo developer) with those of my company. Will keep my work related expectations low and solo developer related high.

Lead by doing. Show what you're doing and why, and get others to support your vision. Then it may be easier to get others do tackle some of the challenges with you.

I did try that and it did not work. See explanation above.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact