Re: svn commit invokes $EDITOR not where svn run from

Alexis wrote:

> What's wrong with a normal fork() + exec()? The child does a chdir()
> and uses probably execvp() on a sprintf() of getenv("EDITOR") and the

Karl wrote:

> If P is also an svn working copy directory, then we can use P's
> temp area. In other words, the target paths should always be

Philip wrote:

> > > If subversion had an interface to /tmp and $TMPDIR (or whatever any
> > > given platform uses) we could use that, such a path should be safe and
> > > so we could avoid the cd. It is likely that such an interface would

cmpilato wrote:

> Windows has environmental variables for answering this question. I
> think the ordering is WINTMP, TMP, TEMP, but I can't recall with
> certainty.

Blimey, I had no idea it would be so complicated :-) Still, I'd like
to try writing a patch - I'm using SVN a lot and this would be a
small payback. I printed HACKING out today :-)

The easiest to implement I guess is use /tmp (Unix) or use
getenv("WINTMP") || getenv("TMP") || ... (Windows). Ok, on Unix
/tmp can be relied upon, but can somebody confirm that the env.var +
fallback env.vars is the way on Windows? Alternatively, Karl's way
daunts me, and I think I would have to chicken out.

Is TMPDIR support in APR imminent? Should I just grit my teeth -
very gently - and type ":r <subdir>/diffs.out" in vi instead of
":r diffs.out". ?

My question then is simply: what is the correct solution? Subject to
how easy it is I'll start trying to code it (and the 'commit
message file not left around on commit error bug' at the same time).

Alexis

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Aug 8 23:53:11 2002