Robert Cummings wrote: > On Mon, 2007-01-08 at 23:59 +0100, Jochem Maas wrote: >> Bill Moran wrote: >>> With reference to: >>> http://bugs.php.net/bug.php?id=40067 >>> >>> I'm confused as to why this was marked bogus, and the message that >>> marked as such doesn't give much insight. >>> >>> It would seem to me that infinite recursion within PHP is a bug. >>> Shouldn't the interpreter catch this sort of thing before it >>> coredumps? Or is the design philosophy for PHP different than >>> that? >> the way I understand it is that there is no decent way for the engine to >> tell the difference between a piece of recursive code that will complete >> and a piece of recursive code that will never complete ... so the only >> sane solution is to leave it up to user land code to make sure the recursive >> loop is not infinite. >> >> put another way if the engine catches/stops what it *thinks* is an infinite loop >> then how would stopping the loop at an arbitrary location be any better >> than dumping core? > > Ummm, on many systems you end up with a core dump with the following > name: core.<pid>. Now multiply that by 1000 at 16 to 32 megs a piece. > Not fun :) core dump files are only generated in debug builds I thought. I may well be completely wrong though :-) > > Anyways, the core dump happens because memory is exhausted and the > engine is unable to allocate anymore. It should be possible for PHP to > exit gracefully upon finding itself out of memory... Possibly it could > allocate another 1k and write to the error logfile on which line it was > running when it ran out of memory. that does sound sane. > Or better yet, maybe a > debug_backtrace() output *drool*. wouldn't the backtrace be massive, leaving you with the same problem as indicated by your point regarding the core dump files? > > Cheers, > Rob. -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
