Re: DAVEM TPF grant#2 report #149,#150,#151

As usual, thank you for the report and work, Dave! :)


On 11/28/2016 10:18 AM, Dave Mitchell wrote:
> (this report covers 3 weeks)
>
> I've gotten a bit behind on logging and reporting my hours recently - in
> particular I spent 4 days at the p5p hackathon, but mainly forgot to log
> any hours.

At first I was surprised, because I know how much time you spent during
the hackathon working and discussing issues, but then I realized that
you are far better at analyzing how much of your time is part of the
grant work. :)

> Anyway, of the logged stuff, I fixed a few tickets and freed up
> some uses of SV flag bits, mainly by eliminating obscure uses.
>
> I've also been looking into whether the lexer's token recogniser
> (keywords.c) could be done in a different way: the current approach, as
> used for the last 10 years, generates a custom code-driven trie, which uses
> very few CPU instructions; but these days branch prediction is more important,
> and that code style kills branch prediction. I suspect that a data-driven
> trie will perform better now.

This is fascinating. Thank you for sharing the details of the work you do.


> [...]
>
> SUMMARY:
>       0:07 [perl #129140] attempting double-free Perl_safesysfree util.c:388
>       3:51 [perl #129991] Assertion failure in S_aassign_copy_common
>       3:21 [perl #129997] heap-use-after-free sv_catpvn_flags() within Perl_do_vop
>       8:44 [perl #130132] Bleadperl breaks Date-Manip
>       1:00 [perl #130188] crash on return from substitution in subroutine
>       8:00 eliminate some ugly SV flag usages
>       2:09 optimise undef assignment
>       4:00 performance of keywords.c
>       1:45 process p5p mailbox
>     ------
>      32:57 TOTAL (HH::MM)

+1!

• DAVEM TPF grant#2 report #149,#150,#151 by Dave Mitchell
• Re: DAVEM TPF grant#2 report #149,#150,#151 by Karl Williamson
• Re: DAVEM TPF grant#2 report #149,#150,#151 by Dave Mitchell
• Re: DAVEM TPF grant#2 report #149,#150,#151 by Sawyer X