Re: [PATCH] Issue #1295

On Thu, 8 May 2003, Blair Zajac wrote:

> Branko ?ibej wrote:
> >
> > Ah, I see the problem, you don't know how memory allocation works in
> > Subversion. :-)

Well, I understand the concept of pools; just not the specifics of APR
pools (yet...). Oh well, I guess this is all just part of the learning
curve.

> > Well, don't worry; apr_hash_make will never return NULL, and we never
> > generate out-of-memory errors.
>
> That's right. Every machine Subversion installs on magically gets an 10
> GByte memory upgrade and can't run out of memory :)

I saw the test for the svn_create_stringbuf() so I figured it was
necessary. I guess the alternative method for handling it would be with
setjmp/longjmp and some kind of exception mechanism or callback.

I see some abort-on-failure logic in the pool support in libsvn_subr.

Is this how it is handled?

BTW: Testing my new patch now. With doc strings, formatting nits, and a
hash to make the algorithm linear.

L8r,
Mark G.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri May 9 05:01:00 2003