Okay, I thought a bit about filling out that survey, but I couldn't because it didn't really fit my needs. After all, it's a Git *user* survey and I'm only a *wannabe* user.
So, why am I only a wannabe Git user (and a frustrated Subversion user) even though I like Git very much? Well, there's a couple of reasons (beware: this rant should have really been a blog post, but I don't have a blog ...; also, being a rant, this post may contain oversimplification, overgeneralization and overexaggeration. Please, don't feel offended and also keep in mind that I'm not a native speaker.).
The most important reason why I'm not using Git is probably that I'm the wrong kind of user: Git is a source code(1) management system for Linux(2) developers(3) and I am
(1) not managing source code,
(2) not using Linux and
(3) not a developer.
Please note that there's nothing actually wrong with that---you can't
drive screws with a hammer and you can't make coffee with a
screwdriver but that doesn't mean that the hammer or the screwdriver
are buggy, it just means you have either the wrong problem or the
wrong tool.
So, I don't really know whether the fact that Git doesn't support my operating environment and my workflow is (a) a problem with Git being too specialized or (b) a problem with me choosing the wrong tool. This is probably actually the most important question I have, because if the answer is (a) I just need to wait some more until Git matures, but if the answer is (b) I need to forget about Git and start looking for another tool.
In my opinion, version control needs to integrate and really blend into the operating environment and the workflow people use, because otherwise, if it *doesn't* integrate but forces them to change their operating environment, their workflow or their habits, people will get lazy. I know I would. And being lazy about version control is even worse than not having any version control at all, just as with backups it gives you a false sense of security.
In my case the operating environment is the Windows Explorer or Eclipse and my workflow is basically dragging and dropping files with the mouse. I can do just that with TortoiseSVN and Subclipse without having to change the way I normally work. With Git I would constantly need to switch back and forth between Explorer or Eclipse and Cygwin. That is, if I actually get Git compiled and up and running *at all*. (Is there any unwritten rule in Git development that says that new features are only accepted if they add a new dependency on a new programming language? Currently Git is implemented in C, POSIX sh, Perl, Python, Tcl/Tk and C++ (if you include QGit) and I've already seen some Ruby code posted to the mailing list.)
Don't get me wrong: Git is very nice. Its similarity to a Unix filesystem and the similarity of its UI to popular Unix command line utilities means that it actually integrates perfectly with the Unix shell, which also means that it integrates perfectly with the workflow of people who mainly work in a Unix shell. I'm sometimes using Git in Linux where I mainly work from the command line and it's absolutely amazing! You have everything you would expect from your regular environment: cat, diff, ls, tar, grep, mv, rm and even fsck!
But that's not the only operating environment, and it's certainly not the one most widely used. In order to be universally usable, Git needs to also integrate with the Windows Explorer, MacOS Finder, Konqueror, Nautilus, Eclipse and Visual Studio (among others). It needs to integrate with project management tools like Trac or Collaboa and in order to do that it probably needs complete high-quality native language bindings for Java, Perl, Python, Ruby, PHP, ...
Okay, let's answer two questions from the survey:
| Q20. What would you most like to see improved about GIT? (features,
| bugs, plugins, documentation, ...)
The most pressing need for me would be a native Windows port. Where "native" means:
- an installer package,
- integration with the Windows Explorer,
- handling of different line-ending encodings and
- addressing the fact that things like SSH clients, diff, patch,
merge, Perl, Python, Tcl/Tk, Bash and so on are usually not part
of a standard Windows installation and are also notoriously
difficult to install.Number 2 would probably be providing the same level of integration for Eclipse.
| Q21. If you want to see GIT more widely used, what do you think we
| could do to make this happen?
What I think is most lacking is the ecosystem *around* Git, which is actually not very surprising, considering that Git is still fairly young compared to e.g. CVS and Subversion. With "ecosystem" I mean things like public repository hosting (currently it's mostly CVS and/or Subversion and some Arch), integration in IDEs (e.g. Eclipse, Visual Studio), project management software (e.g. Trac, Collaboa) and so on.
The general theme is probably: make Git more seamless for more types of users on more operating systems.
I currently use Subversion. I hate Subversion. I constanly hit the limitations of the centralized model, and I hit them *very hard* and *very often*. I would *love* to migrate away as soon as possible. There's only two reasons why I use Subversion: because I can install it with three mouse clicks and because I can use it with three mouse clicks.
Please, make those reasons go away.
Thank you for bearing with me through this long rant.
jwm
PS: I am well aware of the fact that Git is free software and that I
could fix all those issues myself. However, the truth is, I can't.
I'm stupid. I'm not a programmer. So, please don't take this as a
demand or a request, merely a humble suggestion by a humble wannabe
user, whom you are free to ignore.
