Front page | perl.perl5.porters |
Postings from March 2014
On Sun, 02 Mar 2014 07:09:24 +0100, Aristotle Pagaltzis <pagaltzis@gmx.de> wrote: > * Christian Walde <walde.christian@gmail.com> [2014-03-01 21:50]: >> Feel free to bikeshed or merge. :) > > After clarification on IRC about the contentious item I believe you > wanted to say something like this: > > [...] Yes, with the exception that the bullet point needs to point out the worst consequence. The text under each bullet point is merely an explanation to head off the "but whyyyyyy?!" demographic, and i care not much what is actually written there. > (Did I understand correctly that “bugs and exceptions” is what you tried > to say by “crashes”?) As mentioned before (though in slightly different words), bugs, exceptions, crashes. I've seen ALL occur in threaded programs with very little hints as to their origin. On Sun, 02 Mar 2014 14:53:33 +0100, Leon Timmermans <fawaka@gmail.com> wrote: > But shouldn't that give "Invalid value for shared scalar" exceptions > anyway? Christian, can you give a short example of this issue? Actually i can't. I ended up throwing away most thread stuff i ever wrote, but one small piece that actually works. Tried it earlier and i did indeed get such exceptions. Maybe it's because i have a newer threads.pm or i did not accurately replicate what some perl-only modules i used did. To be honest, i'm thinking that maybe it'd be fine to simply remove the third point and rework the fourth to remove the reference to the third. It is the prime example of why this writing has come to be, in that there are issues of varyingly fatal degree present, but nobody knows where they come from or even how to reproduce them. However due to nobody being able to actually authoratively point at a cause, it remains too contentious, and the complexity is sufficiently hidden by the "modules aren't threadsafe" remark in the fourth point, which points to the same consequences. -- With regards, Christian WaldeThread Previous | Thread Next