Re: Linux and -DTHREADS_HAVE_PIDS
[email protected] (Elizabeth Mattijsen)
| Newsgroups | perl.ithreads |
|---|---|
| Message-ID | <p05111b0cbc090a982c37@[192.168.2.109]> |
At 12:47 -0500 12/19/03, Al Tobey wrote: >I was looking at using Thread::Status to work on some problems with one >of my threaded applications, but it requires Thread::Signal, which >currently only works on perls compiled with -DTHREADS_HAVE_PIDS. Actually, the latest Thread::Signal does its own check to find out whether threads have pids or not and does not check Config.pm for -DTHREADS_HAVE_PIDS . >... I'm >running Red Hat Enterprise Server 3.0 with a custom compiled perl, where >I removed that flag because it no longer is true. You mean to say that ./configure didn't detect the -DTHREADS_HAVE_PIDS flag correctly? If that's the case, the simplest way to report this would be to use perlbug. >Two things: Can Thread::Status be made to work without Thread::Signal If you can tell me a way to execute an arbitrary subroutine in a Perl thread, triggered by something outside of that thread, I'd be glad to adapt Thread::Signal to this. If it would have to be in C, that's probably ok as well. Being able to have a subroutine called at will is just _the_ capability you need. Otherwise you'd get into polling schemes such as using an alarm() with a very small value, which would interfere with any other alarms being set, etc. etc. >and would you know whom to contact about making perl properly detect the >THREADS_HAVE_PIDS status? Submit a bug report to p5p with perlbug (a script that comes with any Perl installation). >Almost all modern Linux distributions will have this problem, with the >exception of Slackware (which will probably be there soon). Unfortunately, not many people consider this a problem. I'm glad someone else finds this to be a problem as well. Liz