On Sat, Apr 27, 2002 at 10:16:43AM -0400, Cliff Woolley wrote: > On Sat, 27 Apr 2002, Sander Striker wrote: > > > > http://www.mail-archive.com/[email protected]/msg02492.html > > > > The performance prob goes away when we make apr_pcalloc into a macro. > > Based on your comments, I'm assuming that's a +1 from Sander and Justin, > making three so far. Aaron, I'm guessing you wouldn't object, either. > > Does anyone have an objection? IIRC, the only objection someone had > before was that if apr_palloc() returned NULL, we'd pass NULL to memset > and segfault. But then we got into that same old argument, many of us > (including myself) feeling that just we never need to worry about that.
Indeed that is the only argument I see against this approach, but if we're out of memory, we're hosed anyway. (Because not all malloc() implementations return NULL when EOM.) -- justin
