Front page | perl.perl5.porters |
Postings from August 2011
On 2011-08-16, at 11:02:40 +0100, Dave Mitchell wrote:
> On Mon, Aug 15, 2011 at 09:26:21PM +0200, Marcus Holland-Moritz wrote:
> > > 22:10 [perl #34161] METABUG - (?{...}) and (??{...}) regexp issues
> > > 4:00 [perl #58530] Bus error with constant + overload + stash manipulation + bless ]
> >
> > Just wondering: are those two bugs related (doesn't seem as if they are)
>
> No, they're not.
>
> > and shouldn't the bus error already have been fixed (that's what the
> > ticket is saying)?
>
> Originally the ticket was marked as fixed, but the fix added some TODO
> tests, some of which sometimes SEGV'd, and sometimes passed. (This was all
> before I got involved with the ticket). While trying to fix that, I ended
> up doing a whole bunch of work related to how stashes, GVs and CVs refer
> back to each other (while avoiding leaks and dangling pointers). I kept
> charging that work to the original ticket, although in hindsight I
> probably should have created a new ticket for that work.
>
> That work was mostly finished last year; however, part of my fix
> involved adding backref magic to GVs sometimes. I reckoned I could remove
> this, and spent some time in June and July working on that, before coming
> to the conclusion that I couldn't easily fix it (or more to the point, the
> risk of new breakage outweighed the benefit of the optimisation).
Thanks for the explanation, Dave!
> Here's my complete log of time spent on that ticket:
>
> [perl #58530] Bus error with constant + overload + stash manipulation + bless ]
>
> 2010/06/08 1:00 fix
> Note that although this ticket was closed, the fix
> included some todo tests that might segfault. I've
> been working on making the todo tests pass,
> and providing more general fixes
>
> 2010/06/12 8:10 fix
>
> 2010/06/13 4:50 fix
>
> 2010/07/04 8:00 fix
>
> 2010/07/05 7:00 fix
>
> 2010/07/09 3:40 fix
>
> 2010/07/10 3:00 fix
>
> 2010/07/11 6:15 fix
>
> 2010/07/12 2:55 fix
>
> 2010/07/13 7:05 fix
>
> 2010/07/14 4:50 fix
>
> 2010/07/17 2:45 fix
>
> 2010/07/18 5:20 fix
>
> 2010/07/29 2:35 fix
> make backref magic store single backref directly
> rather than using an array
>
> 2010/07/31 1:15 fix
>
> 2010/08/01 5:35 fix
>
> 2011/01/20 1:00 fix
> add in GvCV_set(P and GvGP_set()
>
> 2011/01/21 2:30 fix
>
> 2011/06/27 3:50 fix
> failed attempt to avoid magic backrefs
> for simple GV <-> CV case
>
> 2011/06/30 1:25 fix
> 2011/07/01 1:40 fix
> 2011/07/04 2:20 fix
Thread Previous