I have prepared the change to remove the pool arguments to
svn_error_create() and friends. Before I commit it, I want to confirm
that everyone is on board with it, since it won't be easy to revert
once other changes go in.
Specifically, Greg Stein raised the objection (in the issue tracker,
not on the mailing list, because he's naughty) that an application
which embeds Subversion will lose the guarantee of errors being
cleaned up when the top-level Subversion pool is destroyed. I don't
think this guarantee is worth having a pool argument every time an
error is created. If everyone does their job and clears errors
they're ignoring, then no memory will be leaked; and if a few bits of
code don't do their job and error memory is leaked, then the leak will
be small. (And in that case, the leak already exists and applies to
applications which never clear the top-level svn pool.)
So, last chance for a -1. I'll wait 24 hours or so.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Nov 10 01:39:27 2002