Re: svn commit: rev 4012 - trunk/subversion/libsvn_client

[Karl Fogel <kf...@newton.ch.collab.net>]

Justin Erenkrantz <je...@apache.org> writes:
> Though it does look like the previous code made a mistake in not
> sending loop_pool to harvest_committables...  -- justin

I have a dim memory that that was on purpose too -- that
harvest_committables() makes a promise about what pool certain
returned objects would be allocated in, and therefore you can't pass
the loop_pool in the recursive call, you have to pass the top pool
through all the allocations.

No, wait, the add_committable() helper function just gets the hash
table's own pool, in fact it doesn't even take a pool parameter.  So
that's not a concern.

You can ignore this mail :-).

But looking at the doc string for both the public function and its
helper

   svn_client__harvest_committables()
   harvest_committables()

...I don't see "POOL" mentioned at all.  Probably should, at least to
say that *COMMITTABLES gets allocated in it.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org