On Tue, Oct 19, 2010 at 04:59:47PM +0200, walter harms wrote: > > > Mikhail Gusarov schrieb: > > Twas brillig at 10:07:23 19.10.2010 UTC+02 when wha...@bfs.de did gyre and > > gimble: > > > > wh> malloc will return (void *) so there is no need to cast. > > wh> i do not know xmalloc() but i see no check for OOM condition. > > wh> instead of removing xmalloc() it would be more useful to follow > > wh> XtMalloc() that has error handling inside. > > > > Sho what's the reasonable error strategy? Shut down the server? > > > > xmalloc is just a funny name for malloc nowadays. > > > > You just run out of mem. What else can you you expect ? > Either you crash with segfault because you are accessing > unrelated memory or you seqfault instandly because you > access NULL. I prefer an errormsg telling the user what the > problem is (OOM) instead let him curse about a sudden crash. > > > I have seen xmalloc() in several unrelated projekts doing exactly > this: return memory or die. (xcalloc etc. doing the same).
I'd argue that none of these code paths are actually tested, so while it is a nice theory, it's likely that trying to print an error message and shutting down the server afterwards will eventually result in a crash anyway. Cheers, Peter _______________________________________________ xorg-devel@lists.x.org: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
