Re:How to Fail at SVN
by pquerna on Wednesday August 25, 2010 @12:49PM
(#1633363)
4/5 are problems unrelated to SVN. Heck, Git can burn you on all of those.
1/5 is about using svnserve instead of file:// and network mounts, and doesn't need to be a rant, just a don't use file:// protocol for shared repositories.
Re:How to Fail at SVN
by rix0r on Wednesday August 25, 2010 @01:23PM
(#1633472)
I'm guessing that you've never had to use svn+ssh:// with a group of developers. The file permissions hell you get into using that with multiple users is very frustrating.
I much prefer the http:// access method, and have never had a locking problem.
Re:How to Fail at SVN
by sapphirecat on Wednesday August 25, 2010 @01:33PM
(#1633502)
> By default in SVN, everyone is committing to the same - and indeed the master, de-facto, controlling, always perfect - branch. Trunk.
It's worse than that; svn asks you to maintain the trunk/tags/branches directory structure yourself, rather than including them as core concepts in the VCS itself(1). So, someone can commit an update to something under tags/ and svn won't even give out a warning, since it has no idea that tags/* are supposed to be write-once.
1) Looks like they added `svn branch`, sometime after the 1.4.x version I'm actually using.
Re:How to Fail at SVN
by DanHulton on Wednesday August 25, 2010 @01:09PM
(#1633429)
Pffft. Complaining about TortoiseSVN? Use SmartSVN - it works in all 3 OSes, too. I know, I've had to use it in all 3 and it's nice to have that consistancy.
Re:How to Fail at SVN
by dasil003 on Wednesday August 25, 2010 @01:41PM
(#1633540)
Every time I read something about SVN I just thank god I don't have to use it anymore. Just the thought of a merge-heavy workflow in svn gives me the willies.
1/5 is about using svnserve instead of file:// and network mounts, and doesn't need to be a rant, just a don't use file:// protocol for shared repositories.
I wish HN had downvotes for an article like this.
While you're at it, the word you wanted was 'administer'.
I don't see how that makes this a particularly bad article, though.
I much prefer the http:// access method, and have never had a locking problem.
It's worse than that; svn asks you to maintain the trunk/tags/branches directory structure yourself, rather than including them as core concepts in the VCS itself(1). So, someone can commit an update to something under tags/ and svn won't even give out a warning, since it has no idea that tags/* are supposed to be write-once.
1) Looks like they added `svn branch`, sometime after the 1.4.x version I'm actually using.