Re: svn_pool_xxx vs apr_pool_xxx

On Fri, Aug 15, 2003 at 03:45:31PM -0400, Greg Hudson wrote:
> On Fri, 2003-08-15 at 15:41, Sander Striker wrote:
> > It's specific to all the clients we produce. If someone uses our libraries,
> > we shouldn't force our abort handler upon them. They may not want that.
>
> You haven't built a case for this assertion. What else would an
> application possibly want? It might want our libraries to be better
> behaved (i.e. to return an error on allocation failure), but it can't
> have that right now. "We produce a SIGSEGV on allocation failure" is
> strictly worse than "we produce a SIGABRT on allocation failure."

Actually, that is "We *may* produce a SIGSEGV on allocation failure, but we
could also behave very, very strangely. Possibly corrupting data. We don't
know, but hey... it is your choice."

I'll take the SIGABRT any day at the *exact* spot where memory was not
available.

Yes, we're a library, and yes, it isn't optimal behavior. But I'm *very*
leery of not having the abort in there.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org