[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn commit: r14643 - in trunk/subversion: include libsvn_client libsvn_ra_dav libsvn_repos libsvn_wc

From: <kfogel_at_collab.net>
Date: 2005-05-09 19:16:23 CEST

Jani Averbach <jaa@jaa.iki.fi> writes:
> I have one concern: We are now allocating 100K of memory in places and
> in function call graphs where we previously did allocate nothing.
> I know that our policy is that caller should take care of called
> functions pool usage and it's pool's life time. However, I am really
> keen to see that in this case we explicitly use subpools inside those
> functions where stack usage of SVN_STREAM_CHUNK_SIZE has replaced by
> pool. The reason is that this change is going to make in 1.2.0 with
> one week soak, and we could later replace those functions to use
> caller's pool is this is a safe and feasible thing to do. I tried,
> but I couldn't check all call graphs and deside that they are safe,
> and we are not looping over those allocations.

I thought about doing that, but my feeling was: we've had a clear pool
usage policy for a long time, and those callers aren't doing what they
ought to be, we need to find out about it. We can fix them in 1.2.1
if necessary.

It would feel wrong to me to put non-standard pool code into 1.2.0
because we're not confident that our upstack pool usage is correct.
If we took that philosophy, there'd be shim code all over Subversion.


To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon May 9 20:01:18 2005

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.