Archived

This topic is now archived and is closed to further replies.

Is Subversion production-ready?

This topic is 4943 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi, i am about to evaluate subversion scm for a commercial game project which currently uses CVS. I read alot of comments from people who use it for their hobby/free projects but not about experiences on reliability in production environments. - Security? ( The website says its possible to tunnel over ssh) - Are the gui frontends stable/useable enough for daily use? (Win32, Linux and MacOSX platforms are used, repository is hosted on linux) Does anyone of you have experience with SVN in serious projects? Thanks for notes and comments on it. regards, Stephan

Share this post


Link to post
Share on other sites
I use it for my project, which isn''t exactly serious, but is medium sized (greater than 10k lines of code). I use TortoiseSVN and AnkhSVN. Between the two I can do everything I need to.

The only issues I ran into were server-side and involved the usual *nix permission and umask gotchas. They were fully documented in the SVN book and I attribute them to my own ineptitude.

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
We use Subversion (on a Windows 2000 server with NTLM authention, under Apache) and TortoiseSVN (on Windows XP clients) for a project with good results. We''re 10 people and it''s a lot of code. It''s a perticular hardware device with many low-level components in the source tree, about 20 MB in the tree. We''re happy with it overall. It''s for an intranet only, so security isn''t really a major concern.

It''s tons better than CVS. I personally prefer Perforce, but with pricing taken into account Subversion wins the race.

We don''t use AnkhSVN. We used it for some time, but it was painfully slow to open a workspace with 10 projects. TortoiseSVN is quite neat and a good enough client.

I haven''t had any stability or other problems with neither Subversion or TortoiseSVN.

Share this post


Link to post
Share on other sites
quote:
Original post by Anonymous Poster
We don''t use AnkhSVN. We used it for some time, but it was painfully slow to open a workspace with 10 projects.


Have you tried it lately? The solution opening performance has been improved substantially(up to 10 times faster in some circumstances) since 0.4.1.

The Subversion solution itself, which has ~80 projects, now opens in 10-12 seconds on my computer.

--
AnkhSVN - A Visual Studio .NET Addin for the Subversion version control system.
[Project site] [Blog] [RSS] [Browse the source] [IRC channel]

Share this post


Link to post
Share on other sites
Thank you very much so far.
Do you also put your assets in there or use a separate versioning system? And is anyone using ssh tunneling using the standalone server?

regards,
Stephan

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
quote:
Original post by Arild Fines
Have you tried it lately?
No, it was months ago. I guess I could give it a second chance

quote:
Original post by stephanh
Do you also put your assets in there or use a separate versioning system?
What do you mean by assets?

Share this post


Link to post
Share on other sites
quote:
Original post by Anonymous Poster
quote:
Original post by stephanh
Do you also put your assets in there or use a separate versioning system?
What do you mean by assets?



Media files, mainly binary data (prised as SVNs advantage over cvs) like models, sounds, textures etc.

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
quote:
Original post by stephanh
Media files, mainly binary data (prised as SVNs advantage over cvs) like models, sounds, textures etc.
Indeeed we do. It works great.

Share this post


Link to post
Share on other sites
How do you handle the non-mergeability of those "assets"? Subversion still lacks locking(but they are debating the implementation now and it looks like it''s definitely going to appear in Subversion 1.1).

Share this post


Link to post
Share on other sites
Whenever there is a conflict in merging (which happens with binary data whenever there are concurrent modifications), Subversion asks the user to resolve the conflict manually.

What you do in this case is really up to you, and a matter for development team policy.

This isn''t really much different from VCS that use locking rather than merging. In both cases, you have to setup a policy for what happens when two developers want to change the resource at the same time (or decide on a case-by-case basis if it doesn''t happen regularly). In both cases, at least one of the developers will get notified about the issue.

We use Subversion for the storage of binary data (textures, models) on a medium sized project, but since each developer has a pretty well-defined working set, we don''t really get conflicts. When they do happen, they get resolved on a case-by-case basis.

cu,
Prefect

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
quote:
Original post by Prefect
...since each developer has a pretty well-defined working set, we don''t really get conflicts. When they do happen, they get resolved on a case-by-case basis.
I copy that. It''s happend for Word documents once or twice, but usually it''s quite small changes that can be easily resolved manually.

Share this post


Link to post
Share on other sites
Side-note: anyone tried Trac? I like how it seems to be a Wiki/SCM tracker in one but I am not completely fond of having to watch the Wiki to make sure random people don''t mess it up.

Share this post


Link to post
Share on other sites
I''m currently in the process of setting up a SVN repos (for an intranet), and so far it seems pretty neat.

One question is, I understand that between revisions, it stores the DIFFERENCE between the files, rather than storing the whole file again. How reliable is this? As far as I can tell, according to the tests I''ve done with a large amount of (mostly) binary data, it seems to handle it okay, but that can only be really told by using it for a longer amount of time in an every day situation.

So, anyone with experience can fill me in?

Also, I used apache to do an import, and then tried it using svnserve, and found svnserve a LOT faster. Is this normal?

Thanks.

- Thomas

Share this post


Link to post
Share on other sites
quote:
Original post by cow_in_the_well
One question is, I understand that between revisions, it stores the DIFFERENCE between the files, rather than storing the whole file again. How reliable is this?


Reliable in what way? I think pretty much *all* revision control systems work this way. CVS certainly does. Storing fulltext for each revision would be prohibitively expensive.

(CVS only stores deltas for text files, though. For files marked as binary you *do* get a fulltext for each revision. This is one of the main reasons why CVS is a bad choice if you want to version a lot of binary data. Subversion uses the same binary diff algorithm on both binary and textual data.)

quote:

Also, I used apache to do an import, and then tried it using svnserve, and found svnserve a LOT faster. Is this normal?


Yes.

--
AnkhSVN - A Visual Studio .NET Addin for the Subversion version control system.
[Project site] [Blog] [RSS] [Browse the source] [IRC channel]

Share this post


Link to post
Share on other sites