Project Hosting + Source control

Generally, what do you do for this /prog/? What is your preferred reversioning system (git, svn, etc)? How many people are members of the big project hosting systems or have their own projects on them (github, sourceforge, google-code, launchpad, etc).

Alternatively, how many host your on your own site, or simply locally (offline).

gay

not sure if ENTERPRISE QUALITY

local git repo

I used git for my own projects for a while, wasn't bad.

google code + svn

Preferred system for personal projects is git, with master hosted locally via dyndns. Last two jobs have used svn hosted on an internal server.

>>1
Personally, I use SVN for my local projects. I don't use any hosting service.

I've tried setting up a server end of git, but that never worked out well on a Windows machine and I find it deplorable that I need a middle man for some form of centralization for git source control. I refuse to use github, et al.

Definitely looking forward to Subversion 2.0. http://subversion.tigris.org/roadmap.html - they have more decentralized options planned for 2.0.

phpbb + edit

>>9
what

Local git repo with trac and a github mirror.

>>8

Subversion 2.0 will never happen. They're already struggling seriously to rewrite the wc library, and that'll hardly give you any new features.

I think Subversion is very misdirected. It might be useful for playing locally and for bigcorps, but goddamn, it's seriously lacking.

Also, I hope you don't think you need to set a git server for anything. I don't think you understand git, maybe subversion broke you forever (don't worry, it's normal).

Use GIT and most of my projects are rails projects so I use Heroku to host my git repo. Private and free!

>>12
Distributed should be able to de-facto support some sort of centralization, yes? What's misguided about this notion?

The guys I work for right now use git (...on github).

I prefer SVN, because there's quite simply better tools for it, especially on Windows.  I don't want to have to have a full cygwin install just to mash some git commands into a shell.  I like actual integration into the Windows shell, like TortoiseSVN.

Git Is Turds

I have tried to use SVN and git for my little one-man projects but I'm somehow too retarded to figure either of them out =/  It's annoying.

>>17
svn is fine..  i found some thing on google ..like "how to setup subversion in 4 minutes"..shit was so cash

>>18
I actually wrote a utility to manage my repositories through svnserve. It can create the service, delete the service, create repos, mirror repos, manage credentials, etc. It even has batch dump/restore/sync repo functions.

It took me about a week of programming with terrible hacks, but it's extremely effective at management.

>>14

All the distributed SCMs support "centralization", I don't think you understand them (or how do you think projects like Firefox work? certainly there's an authoritative central repository, this is no different from whatever you have in mind)

SVN is misguided because they're spending a lot of energy solving the wrong problems. Case in point: the working copy library rewrite. A ton of work, same shit (actually worse so far). What they should have done: checkout mirrors the repository locally, ala distributed. SVN supports all the needed operations directly from the repository (getting files, making diffs and such). The repository is compact on disk, specially compared with the old working copy format (it's very possible that it'll be smaller). But more importantly, you suddenly don't have to get the network for most of the commands, since you have all the history locally. A much better design.

In fact you can simulate it by using the mirroring tool and then checking out your local mirror. Works awesome but it's not practical unless you have some support scripts (you have to commit to the original repo, not your mirror...)

I really don't understand what they're trying to do anymore. It's like they consider anything distributed as pure evil and ignore all the good ideas, rather burrowing themselves in their aging design.

BTW from the look of it I think a lot of people here have the misconception you need some kind of server to work locally with SVN or git. This is completely untrue. SVN can work the repository directly, and with git you are the repository (and if you want to share/publish it, ftp/http is more than enough)

>>20
Sure thing, except I don't want to have to deal with FTP/HTTP. Why can't I run a git server to host my own repositories and still be a mere mortal?

How do I think Firefox works? There's no Firefox git repository, and if there is, it's a git interface to CVS/SVN. What the hell kind of trolling is this?

>>20
Much more would have to change to make svn distributed than you seem to think. One of the basic concepts in svn is the linear, chronological repository-wide revision numbering. You can't have non-linear revision history, and there's no easy way to add it. This is absolutely fundamental to svn, and its creators still firmly believe that this is a feature, not a design flaw.

Keep in mind that the simplest form of non-linear revision history in svn, that is merging branches, is still done in a completely backwards way: you merge the branches into your local copy, fix up any conflicts, then commit a giant patch which is all the branch changes. It appears as one revision in the logs, with at best some merge-tracking metadata and at worst no comment whatsoever regarding its source. Good luck to the poor souls who have to dig through your branch changes later on.

This is the same reason why hosting a local svn mirror does not really work. Access to old history locally is a very minor feature; what you really need is to commit locally and push later on. push/pull is simply not possible unless you manually merge each commit, because the revision history has to stay linear.

Sure, they could replace the numbers with hashes to support non-linear history, fix branch merging to link history properly, and finally support push/pull with mirrors, but then after all that work they'd basically end up with git. What's the point? Besides, you are never going to convince them that non-linear history is a good thing. They're simply not listening.

>>21
You don't have to deal with FTP/HTTP. Hell, you can put the master repo on a Windows shared folder if you like. Security is something totally separate from git itself; if you don't care about security, then all you need to do is make the folder public somehow with read/write access. This is why most simple setups just access it through ssh.

And as everybody keeps saying, you *don't need* a git server *at all*. The git and svn clients are both completely capable of managing your local repository, with *no server whatsoever*. IHBT?

>>23
I don't want a fucking file share or a directory share. I want a god damn service of some kind that allows 24/7 access from anywhere, provided the credentials are right, to a centralized location.

I should not need whatever stupid-ass hacks to achieve this, I should be able to run some daemon. I've attempted to set up such a thing with git, but it proved excessively difficult to configure and set up (with no shortage of hacks).

I don't give a shit about managing local repositories. I want the ability to distribute with a workflow model that I see fit. I can't do that with git. It's either house it locally and fuck you or the alternative, fuck you and use some shitty web service. What is so difficult to understand about this?

>>24
This daemon is called "sshd". It's pretty easy to get it working.

>>24
So what you're basically saying is that every SCM should roll its own shitty server, right? IHBT

>>26
Yes. That is, after all, A BASIC STANDARD FEATURE THAT EVERY OTHER ONE HAS.

Of course, it wouldn't surprise me if you were like the other freetards:
http://stackoverflow.com/questions/233421/hosting-git-repository-in-windows

Most of the responses here boil down to:

1. Why would you want to do that? Switch to Linux, you eediot.
2. Use some other software.
3. Use this shitty shared folder hack.

Instead of telling him what he wanted, they were telling him what he should want or some awful hack that is no substitute for basic, standard functionality. I think there are only about two useful responses in there.

God damn it, that attitude is intolerable. Continue peddling your shit software, freetards. Nobody's going to listen.

>>27
Jesus Fucking Christ, you're far more retarded that I initially thought.

Without having done it before, I just set up a "git server" - to call it something - on Windows, natively (no cygwin or other stuff). It runs over HTTP and has basic username/password authentication. It allows both pushing and pulling.

It took me 20 minutes (including testing). Now I could do it again on 5.

I love when people can't see past their own nose. And no, I doubt Joel's shithole is going to help you.

>>28
Really, Superman? I'm sure you wouldn't mind publishing the method, then.

>>24,27
What the balls. How are you still chirping about this? Just do what everyone has been telling you OVER AND OVER, just fucking use ssh.

You should have remote login enabled already if you aren't in the stone ages. Then just git clone ssh://<yourfuckingbox>.dyndns.com/path/to/your/repo. This should literally be zero work.

If you're still stuck, I don't know what to tell you. You're a retard. (Also, hilarious that you think setting up a cvs or svn server is less work.)

_____________________________________________________________________
/                                                                     \
|         /prog/ Public Service               Announcement            |
|                                                                     |
|     To everyone who uses the terms "freetard", "M$", "Microshaft"   |
|     or equivalent, take your shit elsewhere. No-one here gives a    |
|     crap and you are lowering the already low intellectual barrier. |
|     These terms are the technology world equivalent of comparing    |
\     X to the Nazis. /g/ will welcome your kind, we do not.          /
 ---------------------------------------------------------------------
        \   ^__^
         \  (oo)\_______
            (__)\       )\/\
                ||----w |
                ||     ||

>>24
I should not need whatever stupid-ass hacks to achieve this, I should be able to run some daemon
If you actually read the documentation for these revision control systems (svn, mercurial, darcs, etc), you would find that most of them recommend *not* using the included daemon. They are only meant for testing purposes or for small wholly internal environments, because they are not properly secured. A web server is the correct way to set these up for any outward-facing repository, because their shitty daemons are likely vulnerable to attack. Likewise, ssh is similarly secured, so it is safe to use if you don't care so much about user identification / fine-grained permissions.

(the exception to this is (as always) CVS, which last I recall has the horrifying requirement of running its daemon as root, requiring running it chrooted to protect your system. let's not even start on the problems with CVS.)

>>31
how dare you tell me what to do?! fucking nazi!

>>31
There are many valid reasons for drawing comparisons with Nazis. A lot of otherwise intelligent people have a knee-jerk reaction to dismiss all such comparisons as hyperbole, as if Nazism and fascism were fairy-tale things of such unimaginable evil that they could not possibly really exist on Earth. I have some bad news: they were real, and they could happen again.

>>34
Muslims are the new Nazis. Well, the new Catholics to be exact which (a few centuries ago) were much worse than the Nazis.

>>34
Of course there are valid reasons to draw comparisons to the Nazis, I've done it several times. But it the number of valid comparisons is vastly smaller than the number of bullshit uses to attack anything a person doesn't like.

e.g. Contrasting the way in which modern democracies try to wrangle extra control over their citizens by invoking the terror card, with Hitler's rise to power by using anti-communist rhetoric, may be valid. On the other hand, the vast majority of Nazi comparisons are of the type, "Healthcare is what hitler would have wanted" and, at best, irrelevant, and at worst, an insult to WWII veterans and Holocaust survivors.

>>36
It's true that most comparisons to Nazism or fascism are spurious, but it's also true that any such comparison will draw that kind of criticism, no matter how valid it may be. That needs to stop.

Newer Posts

---

Don't change these.

Name:		Email:
Entire Thread Thread List