I love bitbucket!

nargh it doesnt support SVN, only mercurial and git

we have been using unfuddle.com, but it only is free for 2 users - we had to merge the other stuff by hand
bitbucket is free for up to 5 users, but no svn =/

you know of alternatives ? (has to offer closed source repositories)
we tried git once and didn’t get it
never tried mercurial

WRT: CVS - yeah, single developer (me), single branched…fast connection to repo…so I’ve never needed to muck with any of these fancy no-fangled things.

But you still have the possibility to go back some commits, which is awesome :slight_smile:

To be honest, I don’t get the comments on this thread about Git being obscure or any more difficult than CVS. As far as I can see, it is exactly the same, except when you commit nothing happens on the server (server aka other repository). So you type ‘git push’ in a dos box - one extra step vs CVS. But the benefit is you have local versioning (which admittedly I’m not using, yet).

The point at which you said “git push in a dos box” is where I switched off :wink:

Got to have a GUI that an idiot could use or I won’t touch it.

Cas :slight_smile:

http://code.google.com/p/tortoisegit/

Yeah I know about Tortoise :slight_smile: I just need to get my head around the Eclipse GUI for it really as I rarely use Explorer to do actual source code management.

Cas :slight_smile:

My problem with egit, aka the Eclipse Git is that there’s no context. Don’t know if I’m missing something here, but whenever I commit, it ALWAYS asks for all the files and to select/deselect. When I used SVN, there was context - i.e. if I was in a particular folder, it’d only ask me to select/deselect anything in that folder.

I find egit such a pain when I want to selectively commit a few files when I’ve changed 99999. This is quite realistic - e.g. you’re working on something, you don’t know if it’s stable, but in the meantime a critical bug occurs and you might like to commit a part of your change you know works. Sure, you can branch your change and then backport etc etc, but that wastes lots of time compared to committing 1 file.

With EGit, I believe you can right-click anything under version control and select “add to index”, which is git’s weird terminology for adding it to the changeset. From there you can keep the team window open and actually commit/push from there. Might even be able to do it from that window without the add operation first.

With IDEA, it’s two keystrokes to commit the current file with whatever version control system I’m using. Unfortunately its VCS GUI is pretty deficient in every other respect.

I’ve worked with all the 4 mentioned here (CVS, SVN, hg and git).
The easiest transition was, of course, from CVS-> SVN. We did that, cause SVN is better. Can’t remember why :slight_smile:
Transition from SCN -> hg took some time. We had some minor problems with open tips and the need for regular merge. “Rebase” was the magic wand that fixed everything (also, we got a small presentation by ppl of bigger knowledge)

Now I had to use git, cause I wanted to contribute to a github project. I thought I knew the idea and the concept of dvcs but after using git, at some point I felt like giving up everything. I don’t know, if it was the tooling or the help/tutorials or git having it’s root with linux (I mean the mindset and concepts) but for me, getting git to work and understanding everything was quite a hurdle to overcome.
And then I had to do my first merge with git … -_-

For hg (and git), one of the small nice features is how the verion control files are hold. They are in a single folder that sits in the root, opposite to cvs/scn that have the metadata in every single versioned folder.

In some random version of ToirtoiseSVN that was released about a year ago, they switched to storing all data in one directory - for what it’s worth.

That is defenitly a nice move.

Not so random - it happened in SVN 1.7.

How is 1.7 not random? :slight_smile:

At work we ran into trouble when upgrading to this new ‘minor update’ (because at times directories were shared and we had different SVN plugins installed on the workstations that b0rked on the new/old structure, yeah, errors on so many levels, I know)

Because it’s the next logical sequential number after SVN 1.6? Although you can go into a discussion about versioning numbers there, with such a huge change in the way it works I would have moved it to 2.0 myself. TortoiseSVN is now not so nice to use anymore until you upgrade everything to SVN 1.7 format…

A nice way to learn git is: http://try.github.com/
It takes around 15-20 minutes.

I used to hate git, now is my preferred choice. I keep the repos in dropbox and clone them wherever I am and it works great.

I found no good git integration with eclipse so I just use the command line and “gitk”. Very rarely I need to do some fancy stuff and if I get lost, I use SourceTree http://www.sourcetreeapp.com/.
I’m sure there are good GUIs for Linux and Windows too.

My linux box has just discovered git-cola (on the ubuntu marketplace) that ends any need for me to use the git command line. It’s got everything you could want, a nice diff view and even the ability to commit a few lines from a file…

There’s so much zealousness around version control systems.
Although my first choice is SVN, even CVS works just fine for me; it does what it needs to do.

I’ve dabbled a bit with GIT, but found that Eclipse support was lacking; I had to use the command line to check out one particular project because the Eclipse plugin couldn’t handle it. Integrated tooling is way too important for me to get too adventurous with a part of development that already works for me.

Just curious, what part in particular couldn’t the plugin handle?

[quote]Just curious, what part in particular couldn’t the plugin handle?
[/quote]
It didn’t checkout submodules.