If you haven’t checked out Subversion, you should. It’s a free, open source, powerful revision control system that was built to be a better CVS. It was created with clear design goals and built on top of robust, time-tested technologies. If you’ve been waiting for a better CVS and are ready to roll up your sleeves, skip directly to the ‘Installing Subversion’ section of this article.
Making the Jump to Subversion
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I’m currently rolling out a shiny new Subversion server which will eventually replace our current mish-mash of Perforce, CVS and Visual SourceSafe repositories and databases. Subversion has been the perfect choice; not only is it Free, thanks to it’s reliance on Apache we’ve been able to integrate it perfectly into our Windows 2003 ADS environment. Users are authenticated using mod_pam, which in turn uses Winbind to authenticate via. the ADS ADC. There are a couple of small server-side scripts which create the Subversion access-control files from a series of ADS groups. All of this means that our users can use the Subversion repositories using their Windows username and password, and we only have a single point of administration instead of having to maintain a series of individual usernames, passwords and ACL’s.
The only small downside we’ve run into is the lack of good frontends for our non-technical users. The best so far is RapidSVN but it has a few usability issues E.g. you can add a repository and browse it, but there is no way to simply tell RapidSVN to create a working copy of that repository without opening the Checkout dialog and manually re-entering the URL for the repository.
Still with a little bit of thought and some redesigning of our current Visual SourceSafe databases we’ll be able to move users gently over to this new system and mothball the old data. I can’t wait!
I have looked everywhere, but I can’t find any acceptable backup solution for SVN anywhere.
If it used or if it atleast was possible to use MySQL as backend that would be a non-existing problem as I already have a backup solution for MySQL (online backup).
I know there’s some hacky patch available for it to work with MySQL, but that’s not acceptable either.
So I guess I’ll be stuck with CVS until I can do incremental backups on my files.
And you can’t do checkouts to NFS mounts either which is a requirement here as I have 7 different machines. Having one checkout for each machine is not an option.
A new frontend I’m testing in these days is eSvn, qt based, and looks like it will be THE frontend for linux.
As for windows, I’m using TortoiseSVN and it seems good to me: couple it with WinMerge and a good editor and off you go…
@Magnus:
Backup problems? Why?
Just do a hot-backup, complete or incremental, as described here: http://svnbook.red-bean.com/svnbook/ch05s03.html#svn-ch-5-sect-3.6
All of which with the tools that come with subversion… what problems do you have?
I suggest anyone interested to download and enjoy the free svnbook “Version Control with Subversion” at
http://svnbook.red-bean.com/
Ciao, Renato
http://esvn.umputun.com/forum/viewforum.php?f=2
Forgot to mention it, sorry.
(don’t know if it works on windows: it probably does, but you need all the toolchain for QT, and a license)
It integrates with explorer and it offers a very intuitive way of working with a repository (though for people used to separate apps like SourceSafe it might take some getting used to).
http://tortoisesvn.tigris.org
Tortoise is not an option for the very fact that it integrates into Windows explorer. It looks like Windows Explorer but it certainly does not behave like Windows explorer. All that will do is create confusion. It is also very fiddly to browse repositories with Tortoise, and even harder than RapidSVN to checkout a working copy of the repository you are already browsing. There is no way our current VSS users could switch to Tortoise without training, and even then I suspect we’d have plenty of problems.
Thanks for the information on eSVN As it happens we do have a full commercial Qt licence for Win32 so that could be an option, if the code is clean enough.
Please, try to compile eSvn under win32 and report (mainly to the eSvn forum, but also here would be useful) if it compiles and run cleanly!
As for TortoiseSVN… did you try it or it is just prejudice?
It is really not very different from eSvn or rapidsvn, only you don’t have to launch it explicitly, and all its functions are in the left-click menu.
I learned it in 30 minutes straight (maybe less) do you really have such dumb developers? :-)) I sure hope not! ;-P
Repository checkout: left-click->Checkout
As for checking out while browsing… did you file a bugreport? The devs will probably fix it if you let them know it is a valuable addon feature.
I personally use Tortoise my self for development, and do not find it hard to use at all. There are a couple of things that annoy me a little but other than that I’ve had no problems.
I have just tried to build eSvn on Windows, seems there are a few unix dependancies, but not many. Going to take a look once I’ve some spare time.
Really…what’s with all this talk about switching to it? I know CVS is horribly outdated, but what about GNU Arch? It’s really great…why not switch to that?
Have you actually TRIED arch? It’s not as easy to learn as CVS and is very different.
I know CVS is horribly outdated, but what about GNU Arch? It’s really great…why not switch to that?
—
some people really want a better cvs, not a distributed architecture
i switched to SVN a few weeks back. i like it a lot.
as for backing it up, svn uses standard Berkeley DB files.
unless someone is accessing SVN the files should not be open, and thus can be backed up.
personally if i was going to do it, id tell svn to dump and backup the dumps and not the SVN database.
have a nightly script to create dump files (plain text). and just back them up… shouldnt be too hard.
To Magnus, you cant change the backend.. (personally, Im glad its not backended by mysql. one less server I need to be running). right now I run it as standalone (no apache2). since there is just me.
i also like that svnserver can be standalone too. but im not using that either.
It will be quite a while before anyone here will try to build eSVN on Windows. I’ll have to resource one of our developers to try it out (They’re the ones with the Qt licenses after all). If we ever get around to it I’ll let you know how it goes.
As for TortoiseSVN, yes of course I tried it. My boss tried it. We discussed use cases. It will never fly. It is fine for our developers, who will have no problems using it. It is certainly not good enough for our non-technical users, who are generally just about comfortable with Visual SourceSafe, and only for basic use.
Our Java developers will use the Subversion plugin with Eclipse. The rest of our developers will have the choice of TortoiseSVN, RapidSVN or command line. Our non-technical users will probably end up with RapidSVN unless we can get eSVN evaluated (and we like it).
Success
I have got eSvn to compile on windows. Not really played with it that much to test if it all works. But I’ll try and create some patches for it later.
If anyone is intrested just drop me a email. (help me remember to do it
)
>I have looked everywhere, but I can’t find any acceptable >backup solution for SVN anywhere.
This I cannot understand. Subversion uses the BerkleyDB database engine, which certanly can do backups, hot backups even. A hotbackup.py script is shipped with subversion that does what you want.
svnadmin has the “dump” tool, which backs up a repository
to an ascii coded file and svnadmin restore will .. restore it.
The documentation also have a chapter on backup:
http://svnbook.red-bean.com/svnbook/ch05s03.html#svn-ch-5-sect-3.6
As reported by Mark Hillary on this thread, it seems to work! ^__^
I’d suggest contacting him… and Mark, please let the developer of esvn know, he’ll be pleased, I guess.
I personally contributed a couple of bugreports and the .desktop file for the freedesktop menus, and he added it in no time.
Just out of curiosity: what the “non-tech” users use cvs/svn for? I guess it’s document management, am I right?
If you tested it then I’ll take your word, but to me tortoiseSVN looked actually simpler than the others… you have a green check mark, the file is up-to-date, you have a red exclamation mark, the file has been modified. But then again… svn is very “tool-agnostic”, so to each his own!
> And you can’t do checkouts to NFS mounts either which is a requirement here as I have 7 different machines. Having one checkout for each machine is not an option.
Are you sure you’re reading the docs right? You can’t have the repository on an NFS mount (because BerkDB needs local filesystem semantics, apparently), but I was sure you could check out your working copy to wherever you want.
http://svnbook.red-bean.com/svnbook/ch05s02.html
Uh… you can’t access the repository as local with a NFS mount, but I think you can checkout wherever you want… (otherwise how could it work on a dumb filesytem like fat?)
Just try it!
I’ve emailed Mark so hopefully I’ll be able to get my greasy mits on a copy of eSVN soon.
Yes, our non-developer users include tester, sales, general management and technical authors and yes, it is used for document management. Obviously some of these people are more technically able than others. TortoiseSVN can always be an option for them, if they really want it, but as a fallback we really need something with a more traditional user interface that the other users are going to be more comfortable with. RapidSVN and eSVN are closer to the tools they’re already using (Visual SourceSafe) than TortoiseSVN is.
Even our current CVS users prefer WinCVS or command line to TortoiseCVS. I’ve never asked them why they don’t use TortoiseCVS.
Understood.
I bet you’ll feel a lot better when VSS will be a thing of the past… I’ve read a lot of horror stories on VSS lately. And that’s not even starting to tackle the performance/feature/usability side.
Personally I’ve never used it… and I’m glad about it! 🙂
Read the SVN handbook, backing up is a breeze,
basically everyhting is done trough the svnadmin dump command.
You can backup fully, incrementally from a given revision number delta and so on…
What you have to to is to combine the svm dump with other tools like gzip or bzip2 (or tar) and you are set.
I sure hope a gtk version comes out. I can’t stand qt apps.
Don’t hold your breath… I too prefer gtk apps, but eSvn is good, and at the moment IMHO the best frontend on linux. And that’s saying something, since it is a VERY young project.
I have some hope in the meld project (started as a visual diff tool) that now integrates some svn… but I don’t see it becoming a full featured subversion client in the immediate future.
Let me tell you how svn changed my life, and mental sanity.
My collegues werent used to any kind of files versioning control, that so file merging Caos was a day by day topic.
On the table to be implemented was VSS and source gear. Visual Source Safe its bundled on MSDN universal subscription but repository corruption and non remote access to it makes it from the beggining a dead rat to follow.
Im so glad to had the opportunity to implement subversion.
Its surelly a timesaver and the license/cost issue makes any top decisson maker dont glitche twice.
We mainly use tortoisesvn as our visual svn client, for .net the patch made available cames pretty handy, and apaches2 integration its a true killer combo. There just some annoyances, like file/directory deletion, the merging/diffing tool not easy to work with on big file changes, despite some can use others, and the lack of use of locks(or references to user use), permissive tagging some file as beeing worked by someone would be a must, on a sense that possible conflicts could be avoid from the beggining if you know whos working where. And its only my impression or the explorer shell user feeling its a little slower after installing it?
Im just looking for a svn gui for our macosx designers, svnx is a choice, but im seeing that the lack of login/password makes it not a option if you intend to do commits, since our repos requires it. Suggestions, not cli plz?
congrats for svn developers for such good piece of software
It’s b it useless for me. I’d sooner slam my balls in a fridge door than use a versioning system that requires apache, ugh. Unfortunately, it needs it, and the non-apache server it provides is very simple and doesn’t handle authentication and users properly.
So I think I’ll stick with CVS, it can at least handle these things without needing awful stuff like apache installed and running.
TortoiseSVN DOES slow down a bit explorer.exe, but check if you activated “Indicate folders with changed contents” in the “icon overlay” tab of the settings: as the tooltip says it will sure slow down explorer much more.
Your choice: if you prefer sticking with cvs then you probably don’t need all the features svn has and cvs does not.
Uh… and I think ssh+svn could be authenticated enough… surely more than the standard pserve cvs!
Maybe you are looking for this: How FSFS is Better? Can host on network filesystem. http://web.mit.edu/ghudson/info/fsfs
Atomic commits exist in every source control system I’ve ever heard of (except CVS). Parseable output? Renaming? wow! What killer features, it must have taken them years and years to implement that. Highly advanced.
Also, svn folks mistake “functionality” with usability. The absolute best subversion client I have tried is worse than even the most retarded cvs client. SmartCVS + CVS is about 100x more usable than any svn solution.
And the Apache2 issue is important, it leads to a nasty catch22: nobody remotely serious is going to install apache2 on a production machine. But nobody serious is going to put their repository on anything but a production machine! Oh well :
The whole thing seems to be ‘lets reinvent cvs with all its warts, and slap on a few nice features on top’. I went in expecting some sort of paradigm shift, the kind of ‘aha!’ one feels when switching from emacs to IDEA, from Linux to OSX; a lifechanging experience that made you a better human being. Instead, all I ended up with is the retarded cvs people churning out more crap under a different name. Sub-version is, alas, sub-par in ways that can’t really be remedied without spending another 3 years undoing the evils of the last 3. Ah well, back to unglamorous CVS for me.
Subversion doesn’t require Apache. There is a stand-alone server if you don’t want/need a web-based interface.
Atomic commits exist in every source control system I’ve ever heard of (except CVS). Parseable output? Renaming? wow! What killer features, it must have taken them years and years to implement that. Highly advanced.
You must have been living a sheltered life so far – at work I have to use a commercial source control system which has neither atomic commits nor decently parseable output.
@Anonymous
And the Apache2 issue is important, it leads to a nasty catch22: nobody remotely serious is going to install apache2 on a production machine. But nobody serious is going to put their repository on anything but a production machine! Oh well :
Gee, that’s funny, because almost every new Linux distribution there is ships Apache 2 as the default web server. Additionally, the company I work for has used Apache2 since sometime at the beginning of last year in a heavy production environment without any problems from it. I think you need to re-examine your baseless accusations.
Please read this “Dispelling Subversion FUD” here http://www.red-bean.com/sussman/svn-anti-fud.html and stop saying bullshit about Subversion requirements.
And the Apache2 issue is important, it leads to a nasty catch22: nobody remotely serious is going to install apache2 on a production machine. But nobody serious is going to put their repository on anything but a production machine! Oh well
Oh please…seriously. Do you have a clue how many apache installations there are out there in production, mission-critical applications? More than IIS for sure. Ask Netcraft:
http://news.netcraft.com/archives/web_server_survey.html
<Br> Besides, as many others have pointed out here, apache is a very optional addon to svn.
@renato
its unchecked, i had already spent some time on tortoise settings before, tweaking and trying different ones. The slowleness must be related to icons displaying of course but thats something i need to get with it.
Atomic commits exist in every source control system I’ve ever heard of (except CVS). Parseable output? Renaming? wow! What killer features, it must have taken them years and years to implement that. Highly advanced.
@anonymous
you should read something about svn vs others http://better-scm.berlios.de/comparison/comparison.html before spreading fud and nonsense accusations
> @anonymous
> you should read something about svn vs others http://better-scm.berlios.de/comparison/comparison.html before spreading fud and nonsense accusations
This just confirms that there’s another VCS that’s as retarded as CVS: MS Visual SourceSafe
You know, you think something as boring as VCS wouldn’t generate these kind of trolls, but here we go…
I have looked everywhere, but I can’t find any acceptable backup solution for SVN anywhere.
In the words of Bart Simpson, look harder Homer…
svnadmin hotcopy repo /tmp/repo
tar -C /tmp -jcf repo.tbz repo
rm -r /tmp/repo
If it used or if it atleast was possible to use MySQL as backend that would be a non-existing problem as I already have a backup solution for MySQL (online backup).
It will be possible to use something as heavyweight as MySQL in the future. You will certainly see better performance on small projects with an embedded database. However, when the elusive “backup solution” you’re looking for is a three line shell script, I really have to wonder how sincere your original post is to begin in the first place.
And you can’t do checkouts to NFS mounts either which is a requirement here as I have 7 different machines. Having one checkout for each machine is not an option.
Wrong. What the hell gave you that impression?
Of course, that’s all fine and dandy compared to this:
What’s so special about Subversion?
It’s a mature, highly used VCS with distributed versioning through svk and integration with a whole slew of development tools including Visual Studio.NET. It also integrates with Windows Explorer through TortoiseSVN.
Really…what’s with all this talk about switching to it? I know CVS is horribly outdated, but what about GNU Arch?
Arch is broached on virtually every Subversion related thread on OSnews, by zealots much like yourself who like to tout how “great” it is without mentioning anything specific. Honestly, I really have to wonder how many of you have actually tried both Subversion and Arch on real world projects, considering you have no specific issues to argue with Subversion, and instead just want to babble about how wonderful Arch is, much like this:
It’s really great…why not switch to that?
Let me begin by waking you up to the fact that some of us don’t live off exclusively in *IX neverneverland and actually care about developing projects which can be used on the majority of desktop computer systems in the planet. Using Arch on Windows is arcane and tedious, especially compared to the *excellent* integration Subversion provides through TortoiseSVN and AnkhSVN.
The Arch toolchain is small and paltry compared to that of Subversion. Subversion has an excellent web-based repository viewer through WebSVN, and Subversion archives can be browsed using any WebDAV client, including Windows Explorer.
Given the difficulty of getting all developers on a project to install a 3rd party (i.e. non-CVS) VCS in the first place, I prefer one with more support and better distribution across a larger number of platforms. Care to point me to a Solaris package for Arch on sunfreeware.com? I’m afraid there isn’t one, but there is, however, a package for Subversion:
http://sunfreeware.com/programlistsparc8.html#subversion
The same goes for Fink and other packaging systems. Unless you’re targeting Linux exclusively, and all your developers are using Arch friendly distributions, forcing developers to use Arch is a needless headache for them. After all, wouldn’t you rather your developers were developing as opposed to trying to compile a VCS?
I’d rather use a VCS designed as a generalized CVS replacement as opposed to a VCS specifically designed to replace BitKeeper for Linux kernel devleopment because some GNU zealots don’t like the idea of commercial software (or ‘proprietary software’ in your precious GNU semantics) being used to develop the Linux kernel.
> If you haven’t checked out Subversion, you should.
I haven’t and won’t use it unless I absolutely have to. It’s better than CVS, but why should I go back to it if there are darcs and arch around?
> It’s a free, open source, powerful revision control system that was built to be a better CVS.
Yep, and that’s the problem. It’s just like CVS, a server-client design, heck, it doesn’t even allow for distributed branching. Granted, it does many things better than CVS, but why keeps everyone bellowing that it’s the best?
*sigh*
It’s just like CVS, a server-client design
Which is… just like Arch in many ways… do you have a specific complaint or are you just trolling?
heck, it doesn’t even allow for distributed branching.
Wrong. svk.
Please let me know what project you’re working on which requires distributed versioning. I’d *really* like to know.
Yes, this is the one feature which Arch has worked on pro-actively for the reasons I have stated in my last post. Meanwhile small goals like portability and developing easy-to-use GUI clients or even just prepackaged clients for a variety of platforms have gone by the wayside.
Distributed versioning is really unnecessary except for extremely large projects in which there are multiple development trees being maintained by multiple individuals with portions which need to be periodically merged. At that point, why would you use Arch as opposed to BitKeeper or Perforce? As I stated in my last post, unless you’re developing exclusively for Arch friendly Linux distributions, forcing your developers to build Arch to even work on your project is cumbersome and impractical.
Granted, it does many things better than CVS, but why keeps everyone bellowing that it’s the best?
Support support support support support. Please see my last post to this effect. Subversion is a *practical* VCS to use on a cross-platform development project. Arch is *impractical*, especially if you are targeting Windows.
> It’s a mature, highly used
It may be mature, but _highly_ used?
> VCS with distributed versioning through svk
So you need a special tool to work distributed with svn, something one’d excpect the VCS to do natively.
> and integration with a whole slew of development tools including Visual Studio.NET. It also integrates with Windows Explorer through TortoiseSVN.
That’s currently the only point to use svn over other tools.
> Using Arch on Windows is arcane and tedious
That Windows has no decent shell built-in is not Arch’s problem.
> The Arch toolchain is small and paltry compared to that of Subversion. Subversion has an excellent web-based repository viewer through WebSVN,
Care to give any more examples? Almost every VCS out there has a web-based viewer and AFAIK arch has WebDAV support too.
> Unless you’re targeting Linux exclusively, and all your developers are using Arch friendly distributions, forcing developers to use Arch is a needless headache for them.
FreeBSD has a tla port too, so it can hardly be exclusively for Linux.
> I’d rather use a VCS designed as a generalized CVS replacement as opposed to
> a VCS specifically designed to replace BitKeeper for Linux kernel devleopment because some GNU zealots don’t like the idea of commercial software (or ‘proprietary software’ in your precious GNU semantics) being used to develop the Linux kernel.
How come svn zealots have to think that way? Just because arch was in discussion to replace bk for linux (just as monotone was discussed to replace CVS for the gcc crowd) hardly means it was designed to do _that_.
BTW, in “GNU semantics” BitKeeper would be non-free (even free software can be proprietary — someone holds the copyright on it unless it’s in the public domain — so the opposite of free software has to be non-free software)
> Which is… just like Arch in many ways… do you have a specific complaint or are you just trolling?
I’m not an arch zealot (if mistook me for one), I’m a darcs zealot (darcs knows nothing like a dedicated repository server, that’d be just a working directory served by http).
> At that point, why would you use Arch as opposed to BitKeeper or Perforce?
Because it’s free.
> Distributed versioning is really unnecessary except for extremely large projects in which there are multiple development trees being maintained by multiple individuals with portions which need to be periodically merged.
I don’t get your point. I work on multiple computers, some of which have no connection between them (or it’s highly expensive), and I still want full access to my repositories. Either I don’t get what you’re saying, but I’m using distributed versioning often.
> As I stated in my last post, unless you’re developing exclusively for Arch friendly Linux distributions, forcing your developers to build Arch to even work on your project is cumbersome and impractical.
As I already said in my previous post, FreeBSD has a port for arch, so it can hardly be linux-specific. And I’ve yet found no distro since I know arch that doesn’t distribute it (however I haven’t seen RedHat, Fedora, SuSE, Mandrake, etc. for more than a year, so I don’t know about them).
> Arch is *impractical*, especially if you are targeting Windows.
That may be. As I don’t use Windows myself, I can’t comment on this.
It may be mature, but _highly_ used?
Not when compared to CVS, but it’s certainly an order of magnitude above arch:
Google for “subversion repository”: 27,000 results
Google for “arch repository”: 7,300 results
So you need a special tool to work distributed with svn, something one’d excpect the VCS to do natively.
Subversion is a already a toolchain of modular components. And why would “one” (GNU zealot) expect a VCS to support distributed development per default? Believe it or not most development projects aren’t as large as oh, say, the Linux kernel.
That’s currently the only point to use svn over other tools.
Because it’s practical and supported, as opposed to not requiring a separate tool for a feature useful to 0.1% of its intended users? What an amazing reason.
That Windows has no decent shell built-in is not Arch’s problem.
Lack of practical Windows support for Arch is a problem for anyone considering using Arch over Subversion for cross-platform projects. Arch is therefore impractical to anyone developing for Windows. Meanwhile you’re bitching that Subversion uses a third party tool for distributed development. Not that distributed development is even remotely impractical with Subversion… you just want to tout the one feature of Arch which even makes it worth mentioning.
So in order to gain distributed development support, useful to 0.1% of the projects in the world, it’s necessary to shut out 95% of the desktop computers in the world. Sounds like a worthwhile exchange to me… in GNU zealot logic.
> The Arch toolchain is small and paltry compared to that of Subversion. Subversion has an excellent web-based repository viewer through WebSVN,
Care to give any more examples?
I’d ask you to do the same. Care to point out a web repository viewer for Arch with the feature set of WebSVN?
> Unless you’re targeting Linux exclusively, and all your developers are using Arch friendly distributions, forcing developers to use Arch is a needless headache for them.
FreeBSD has a tla port too, so it can hardly be exclusively for Linux.
Okay, point me to Arch packages for:
* Windows
* MacOS X
* Solaris/SPARC
* Irix
* HP-UX
* AIX
And I’ll be happy.
How come svn zealots have to think that way? Just because arch was in discussion to replace bk for linux (just as monotone was discussed to replace CVS for the gcc crowd) hardly means it was designed to do _that_.
You’re right, GNU also needed to create a GPL counterpart to the GPL incompatible Apache-licensed Subversion. But don’t blame GPL incompatibility on the GPL itself, blame it on the horrorific Apache license with its ridiculous clause asking you to ask permission before using the project name for a derivative. What a horrible clause, seriously.
Meanwhile, back in the real world, Subversion is the most practical CVS replacement out there. It’s easy to install and deploy on a variety of platforms, most notably Windows, it’s mature, and it works.
*kshhh* *kshhh*
In nomine patris et filii et spiritus sancti, begone, troll!
CVS rules, everything else drools! Long live CVS! Screw the rest!
It’s already been mentioned once, in passing, but VS.net 2003 developers might like AnkhSVN (http://ankhsvn.tigris.org/)… I’ve been using it for all my projects since around March and despite a few hiccups and bumps (none of it destructive – I haven’t lost a single piece of code to it), it works very very well for beta software.
The hiccups and bumps are often just me not pressing a couple of buttons in the right order (it’s simplicity fails it sometimes), or forgetting to add files before committing them… but it’s worth a try if you’re looking for free version control in VS.
The google fight mentioned earlier:
Google for “subversion repository”: 27,000 results
Google for “arch repository”: 7,300 results
It’s not really accurate since people often refer to subversion repositories as “svn repository”.
There was another 15900 results for the query: “svn repository” -“subversion repository”
i cant/wont use subversion until i can have the repository on an nfs mounted drive
> You’re right, GNU also needed to create a GPL counterpart to the GPL incompatible Apache-licensed Subversion.
Assuming you’re talking about tla. Tla wasn’t started as a part of GNU, in fact, it only became a GNU project with version 1.0.
Sunbversion is certainly good, but CVS is still going to be the mainstay version control system for many years to come. There’s just a lot of people and projects using it.
Works great. The comments about it causing confusion are completely unfounded – integration on windows with explorer is more productive than using 2 file managers, and also less confusing for newbies who can spend a lot of time checking files side by side. These anti-tortoise folks should get a clue.
Ben: i cant/wont use subversion until i can have the repository on an nfs mounted drive
Version 1.1 (in beta now) will have the option to use a flat file system, which can be mounted through NFS.
Keep an eye on it 😉
I’m always surpised that no one mentions FreeVCS ( http://www.thensle.de/ )when versioning control systems are discussed.
I am by no means an expert on VC systems, but for small dev teams/single developers, it has always done the job in a fairly friendly manner. It’s now open source, and the next version’s going to be call JediVCS evidently.
Where can my company get commercial support for CVS or SVN? Looks like we’ll end up going with Perforce.
Ryan Gordon (“icculus”) used Subversion in place of CVS and Perforce, for some UT2004 work (if I recall right). He found a whole bunch of pros and cons. Scroll to the end of:
http://icculus.org/cgi-bin/finger/finger.pl?user=icculus&date=2004-…
I try to avoid sites like OSNews and Slashdot because I often go away feeling depressed. I read through some of these comments and it amazes me how strongly some of you write for or againstp some of these open source projects. I always thought and hoefully will continue to think that software was created to make our lives easier. I hear so many people bash on something when their criticism often sounds like they know very little about the topic at hand. For those of you who have criticized Subversion I am curious, how many of you have actually tried using subversion for a decent amount of time? I will be willing to wager that very few of you have ever even used it.
I have used subversion for almost a year now and it has made my life easier, there may be a better tool out there but I see no compelling reason to search for it when subversion already does everything I have ever wanted it to do. Granted I am not a super user and my requirements for a version control system are pretty simple but for what it’s worth in my experiance SVN has been a pleasure to use. I would encourage anyone who needs VCS to at least try subversion, if it works for you then that’s great; if it doesn’t then that’s ok too, use what does. Let’s try and give the developers who have so kindly donated their knowledge and time some respect.
Actually, I was hugely surprised when around message #30 we were having an adult and useful exchange of views ON TOPIC with only a passing troll about arch… the rest of the thread has re-estabilished the balance of the universe, I guess.
Just as you can’t mention gnome without a kde troll (and vice-versa, often), or linux without a windows troll, you just can’t mention subversion without an argh troll.
Welcome to the internet community! 😉
Works great.
Never said it didn’t work.
The comments about it causing confusion are completely unfounded…less confusing for newbies who can spend a lot of time checking files side by side.
Unfounded? Maybe you should come talk to some of my users. What I’ve said is not unfounded just because you have decided it is unfounded. I know my users, they won’t use TortoiseSVN. Less confusing? The non-technical users won’t be doing diffs and large commits, they’ll be working with a single file at a time.
These anti-tortoise folks should get a clue.
Whatever Kon. Some of us are working in the real world where we have to roll out Subversion to a commercial production enviroment and support non-technical users. If you’re a TortoiseSVN fanboy then good for you, but don’t come hang around my servers and tell me what software is good for my users and what isn’t.
In our experience (having just rolled out svn in our company to replace VSS) the developers have absolutely loved svn especially when compared to VSS.
However for the non-techs, we are leaving them with VSS as although TortoiseSVN works great for them, the majority of their documents are binary files. While SVN works fine with binary files (storing binary diffs), the non-techs are used to using VSS as a file-locking solutions. Merges are not possible with binary files, and as such SVN (until it implements the planned “lock” paradigm) is not recommended for our non-tech projects (e.g. design).
That’s a good point, but luckily for us it isn’t as much of an issue. We’re a fairly small outfit with lots of different VSS databases, so there is usually only one person ever updating files at any one time. Obviously our development databases are far more active, but we don’t need to worry about merging binary files there.
Your right that it’s an important consideration if you’re going to roll out Subversion and rely on lock-modify-unlock semantics.
@QuantumG:
For commercial support, you might want to check out:
http://www.collab.net/subversion/index.xhtml
CollabNet has a list of SVN consultants that might provide just what you need.
@Vanders:
Locking support is planned for version 1.2, development on that is already being done as we speak. Check the SVN mailing lists if you don’t believe me 🙂
No I really do believe you, but like I said lack of locking in the current release of Subversion isn’t an issue for us anyway at the moment. Lack of locking hasn’t stopped me rolling out Subversion at all.
Subversion may be the greatest thing since sliced bread, but as long as there is no built in support for it in common IDEs it will not be widely used.
We contacted Collab.net for commercial SVN support. They only supply it as part of their complete “solution” most of which we don’t have any use for. I can understand there being so few people supplying commercial SVN support, but where are all the people supplying commercial CVS support? One would think there would be hundreds of them with some with strong reputations. Anyone?