Eric Gillespie <epg@pretzelnet.org> writes:
> > last one proposed by Mike is the one I'd always assumed we'd be
> > using).
>
> That last one is a hack. The problem is that you are using $PWD
> as a place to store tmp files rather than the prescribed location
> for that purpose ($TMPDIR or /tmp on unix, c:\temp if my rusty
> dos/windows memory serves). You could hack around this symptom,
> or you could fix the problem. If that means adding a couple
> routines to apr, that's a win for everyone using apr IMNSHO.
> Finding the appropriate place to store temporary files is a
> general problem portable applications could use help solving.
Actually, I'm gonna have to argue with you here. Karl and I both
believe that one day we will support exclusion lists for import
(similar to the example I gave in one of the "other" solution ideas
for #964. By providing a way for svn_client_import() to ask the
question: "Is there a commit message tmpfile I need to skip?", and
then passing that file as the only item in a genericized exclusion
list to the import helpers, we get:
- the ability to import /tmp from /tmp if the user wants to do
that.
- underlying support for a function that we can later expose at the
client level with extreme ease.
As far as I can tell, import is the only operation that doesn't use a
.svn/tmp/ area for its log message tmpfiles (because, duh, it doesn't
have a .svn area -- it's an import). I don't know if that matters;
just throwing info into the mix.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Nov 13 01:09:02 2002