RE (tilly) 1: Poor man's setrlimit

Very poor man's model you mean. It won't help if you do something like any of the following three lines:

push @big, 1 while 1;
$foo = $bar = "BIGBIG"; while ($foo =~ s/BIG/$bar/) {}
sub oops {my $var = shift; oops($var);}
[download]

I have seen all three mistakes made. In fact just yesterday I discovered an example of the third mistake in the 5.6.* implementation of Carp.pm. But you knew that. :-)

Comment on RE (tilly) 1: Poor man's setrlimit Download Code

Replies are listed 'Best First'.
RE: RE (tilly) 1: Poor man's setrlimit by merlyn (Sage) on Oct 17, 2000 at 00:46 UTC
What makes you think those won't be interrupted? alarm is a signals. Signals happen now. One of the discussions is to make signals "safe", meaning they wouldn't interrupt a Perl opcode, and your example is correct. But for now, they are dangerous, and poof! You be gone. {grin} -- Randal L. Schwartz, Perl hacker	[reply]
RE: RE: RE (tilly) 1: Poor man's setrlimit by tilly (Archbishop) on Oct 17, 2000 at 00:53 UTC
Um, 30 seconds is waaay longer than I would want to run one of those on a box with no limits and nothing in place to handle resource starvation gracefully... Trust me, I am not sitting at the counter now. This is one I have definitely been there on. :-(Why, tell me why, does Linux seem to think that klogd is a good target?)-:	[reply]