Re: Why is "svn log" a network operation

Blair Zajac <blair@orcaware.com> writes:
> It seems that svn log should ideally be a local operation and not
> one that connects to the svn server, for several reasons (I just
> tailed my Apache access log and svn log had hits show up there).
 
But why would one want to do this? That's a lot of data to store
locally, and it can get out-of-date (since log messages can be changed
after commit, and are not versioned). And the only way to see *new*
log messages would still be to connect to the server...

> 1) This would be needed for disconnected operation.

We don't support full disconnected operation. We support certain very
common operations locally ("show me my local changes") but that's all.

> 2) It's faster.

No faster than running "svn log" once and saving it to a file, when
you need it. You can spend the time up front, or you can spend it on
demand. :-)

> 3) If you are working on revision X and you do svn log, then you
> get the log for X+N, which is not what you would expect.

This has nothing to do with "svn log" being a local operation. It is
about changing the default behavior of "svn log" to show a more
restricted set of log messages. This might be good, depending on the
specific behavior.

The phrase "working on revision X" isn't very meaningful, in our
working copy's context. There are lots of revisions lying around (and
anyway, why would the most recent one be of more interest? I usually
want the full log on a file, for example.)

> 4) Cvs does not need a network connection to do cvs log.

CVS does contact the repository to do "cvs log", not sure what made
you think it doesn't (?).

-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:37:08 2006