Front page | perl.perl5.porters |
Postings from December 2013
On Wed, Oct 30, 2013 at 11:29 PM, Father Chrysostomos via RT <perlbug-comment@perl.org> wrote: > On Mon Oct 28 22:02:51 2013, sprout wrote: >> On Mon Sep 16 23:53:11 2013, sprout wrote: >> > How about having an array shared between threads and indexed by cop_seq? >> > >> > Or, if that makes the array too long, extend the struct for dbstate ops >> > and store a separate sequence number. >> > >> > When the op is freed, it can invalidate its entry. We would need three >> > values: 0 = no op available, 1 = no breakpoint, 2 = breakpoint set. >> > >> > The numeric values stored in @DB::dbline elements would be indices into >> > that breakpoint array. >> >> Please review the attached patch, which is also on the sprout/dbline branch. > > In particular, does B::C need to access the extra fields that dbstate ops now have, or can we keep the cop–dbop distinction private? Sorry, just saw this today. No, B::C doesn't need to access private dbstate fields, since this is an unsupported and unlikely usecase. One doesn't dump being-debugged code into C, only fresh code. However, Enbugger has a problem. A dummy nextstate field would be better to have the same memory layout. -- Reini Urban http://cpanel.net/ http://www.perl-compiler.org/Thread Previous | Thread Next