>> setting SVN-REVISION to the repository version (post-commits) would
>> also be really handy.
>
> Why? So that the svn logfile can write "username commit REVNUM"
> instead of "username commit"?
Actually it wasn't the commits I was thinking about. It was to save
having to use timestamps to correlate the svn log info with the Apache
log. (The comment about post-commits was a red herring, sorry I
shouldn't have mentioned it). More specifically, the revision of the
operation may not be the same as HEAD at the time, eg "svn co -rXXX", so
trying to match on timestamp wouldn't work. (Sorry, I didn't make it
obvious, in fact probably made it very unobvious, that I was imagining this)
To attempt to come up with an example use case - which will I think turn
out to have fishhooks in it, say I commit rev 1300 then follow it up
with hasty patch in rev 1301, then I want to find out if anyone has read
that bad revision. It would be nice to know if anyone has read that bad
revision : something like grep "rev 1300" <apache/logs/svn
However in thinking that one through, it occurs to me that not every
operation to log is going to work off a single revision number.
Possibly a commit, in fact, is one of the few that always would
(ironically it would also always be HEAD).
One unsubtle example - SVN CAT file1.c@1234 file2.c@5678 (I think this
works). Then there's diff, merge, copy etc.
So maybe this needs to come down to file-level logging - ie user X
fetched file Y at revision Z. Much bigger problem than your patch was
addressing.
- Walter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Aug 9 04:21:40 2005