Lyle Hopkins <[EMAIL PROTECTED]> writes: >Quoth [EMAIL PROTECTED] (Ben Morrow): > >>This is *not* a p5p problem. These are *not* things that should be in >>core perl. If you want to talk to ActiveState and see if you can >>persuade them to produce a 'perl for incompetent IIS users' >>distribution, which includes things useful to people like that, then >>feel free to. > >Stating that this is not a perl problem, when if perl is ran with bad code >that it can take down a web server, would seem a bit ignorant. Like Joe >said - 'if your roof is leaking you cannot keep blaming the rain'. Would you >suggest that the Perl distribution be renamed to 'Perl - only for advanced >users, beginners not welcome' Or maybe 'Perl - add this to your web server >only if you know how to set your machine specifically to cope with runaway >perl processes' Or even 'Perl - not for incompetent IIS users or users of >early versions of IIS'? I'm sure that would greatly increases Perl's >popularity on windows servers. Or you could have 'Perl - includes features >so that badly coded scripts cannot take down your server'.
I think I agree with you for the opposite reason - I don't want perl neutered to stop it being a real large scale general purpose programming tool. I have written 100,000s lines of perl but never put a single line of perl on any web server. Allowing untested / constrained perl code to be run on a web server is a daft as letting C code run. So perhaps we should build a new executable webperl.exe which is really "safe". A magic command line which which limits CPU time isn't going to help if program is reckless in other ways - e.g. a blocked process consumes no CPU at all.
