APR
23
2006

svn is slow...

svn is slow.
Ever tried svn commit or cleanup over trunk KDE ?
I started a commit, pressed Ctrl-C to cancel it. It didn't cancel within maybe half a minute.
I killed it with -9.
Then I tried update, but it told me that it was locked and I should do a cleanup. So I did a cleanup. It took approx. 5 minutes :-/
Now I'm updating again...

cvs wasn't so bad after all. It felt much faster, and was dead-simple on the server. Shorter urls like "cvs co kdebase" and you had everything, no long "svn co https://whatever..." I never can remember.

Anyway, back to coding...

Alex

Comments

I also can't *remember* the svn co URLs.... So what I often do is "svn info" inside one existing repository, and guess from its 'URL:'-line what the real path might be that I'm looking for.

But you're right: svn operations are definitely slower than cvs ones. Maybe this is because it also allows us for "offline working" and therefore has more overhead for its operation ?


By Kurt Pf. at Sun, 04/23/2006 - 13:04

And did you forgot about the CVSROOT ?

I don't see too much difference between https://whatever and :pserver:[email protected]:/cvsroot :-)


By faugusztin at Sun, 04/23/2006 - 13:46

I found the same thing when working with a project with alot of history.. Using the standard BDB store method makes working with the repo a long process, I found using the FSFS backend make things feel faster.


By tomasu at Sun, 04/23/2006 - 17:23

We already use the file system backend. But I think I remember that we have something like 60 GB there. One of the other things that slows down checkouts a lot is that everything is going uncompressed over https. WIth CVS the files were run through gzip for transmission.


By Scott Wheeler at Mon, 04/24/2006 - 12:43