Hacker News new | past | comments | ask | show | jobs | submit login

I'm the only maintainer of this project. I find revision control redundant, since it takes extra steps to set up and use while being of very little value. I'm reasonably good at maintaining my tree without needing any fancy revision control systems (git/svn/whatever).

I will get hated for what I've just said.




It might provide little direct value to your development process, but the real question is, why are you posting this project on the internet?

Do you think there is a chance it will be useful to others, and that they might contribute back patches that you will find useful? If so, it makes sense to make that as easy as possible. Otherwise you are sabotaging your own goals and putting unnecessary barriers between you and your potential contributors.

Sign-up aside, all you would need to do is type "git init" and then set up a one-line script that will do a commit and push it to github. If you want to be nice, you'll type in a message. Voila, you now have a way to:

* let people subscribe to your project

* automatically maintain and announce your changelog

* let them fork your project, make changes, and send those changes back to you as a diff which you can comment on online

* merge those changes back into your tree with a single click / command

Additionally, while you might be confident managing a single code tree, there are things that filesystems don't do well, like branching. I find being able to try new things and setting them aside for a while is invaluable. It lets me be bold and work on whatever I fancy, without having to worry about what I was doing before.

Don't think of git as revision control, think of it as a little robot that lets you do controlled operations on code trees, like you can with files.


I do the same, sort of. I just use RCS on a local machine. Who cares what other people think?

Any chance we could get a glimpse of this booting? Maybe a YouTube video?

Is it easy to do static linking? I'm quite happy without shared libraries. And I don't mind being hated for saying so.


This is Hacker News; we try to stay civil, so I don't think many people are going to "hate" you for not using (understanding?) version control.

Then again, I don't think many people will take you or your project seriously after this, either. Many of us went past folders and tarballs for version control back in 1992, and personally, there's no way I'm going to touch an OS Kernel that is distributed in a zip file. That is not how you do open source -- for what it's worth, you may want to keep your tarballs to yourself: that'd serve the same practical purpose.


I'm going to take christina_b seriously because this is a seriously complex problem, and it's impressive that she's gotten so far with it.

I don't expect that any of the people saying "you should use github to get contributors" actually have the wherewithal to contribute.

Moreover, while I disagree with her stance about SCM, I find it incredibly bikeshed-y that people are talking about that instead of about the technical details of how such a thing is implemented.

Don't any of you have questions or comments about this other than cheerleading, or a bikeshed about SCM? I posted some similar projects that have taken different approaches.


> there's no way I'm going to touch an OS Kernel that is distributed in a zip file

Okay.


That makes sense if it's just you, but you must be losing a lot of potential contributors by not being on Github, let alone not using any version control.

Are you just not that interested in contributions from the public? If so, I think it's possible you're greatly underestimating community interest in the project.


The choice of SCM is the least important factor when it comes to contributors to a genuinely complex problem like this one.


I find I'm always happy I used hg when I get the inevitable sneaky bug in unrelated code. Being able to easily step back through history and run my tests at every step to see where they fail has been very useful many times.


I use hg locally on personal projects also. I very rarely actually need to go back to a prior version, but it gives me more confidence to try different things, knowing I can easily revert to a working version.


> I find revision control redundant, since it takes extra steps to set up

Such as typing "git init"?


And starting a series of steps that have been automated for you?

For a one person project, git is unneeded overhead.


What series of steps? git push?

> For a one person project, git is unneeded overhead.

You've never heard of git bisect?


I hate you! More seriously git is super easy to begin to use, and even more so for small projects and / or projects with a small number of contributors.

Setup? git init; git add . (if no unwanted binary in the tree) Something interesting has been done? git commit -am "rock more" New source file? git add file; git commit -m "new foobar module"

What value does it bring? Great revert capability, bisecting bugs, and so over. I agree most of the time I don't need it, but well, when I do it is handy. It is like being insured.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: