Re: [PATCH] Fix memory leak like situation when parsing a svn_config_t object repeatedly

On Sun, 2005-07-24 at 10:41 +0200, Bernd Rinn wrote:
> > The standard workaround for having to use a baton pool instead of a
> > passed-in pool is to use a subpool of the baton pool, yes. Since we're
> > iterating here, it isn't even terribly awkward. But yes, the API should
> > include a pool, although it's not terribly important.
>
> Maybe I explain the use case so that you can see it is a useful one:

I wasn't disputing the usefulness of your use case; the "workaround" is
for us, because we have a broken API, not you.

I've committed r15404 with the workaround. I'll leave it to someone
else to code up the better long-term fix of revving the enumerate APIs.

Although r15404 is suitable for backporting, I haven't proposed it for
any previous revs because only a small amount of memory is leaked per
call, and that generally isn't a problem. Since you're using APIs which
postdate 1.2.x, I'm guessing you don't particularly need a backport, but
if you do want one, speak up.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Jul 24 18:29:54 2005