Re: DAVEM TPF Grant#3 June 2026 report
[email protected] ("Core Grant Manager") Sat, 18 Jul 2026 15:13:45 -0400
| Newsgroups | perl.perl5.porters |
|---|---|
| Message-ID | <[email protected]> |
--eda9bc239f817ba1db5a58e46d12bafcbfaa6ddd
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
On Tue, Jul 14, 2026, at 7:20 AM, Dave Mitchell wrote:
> Last month was mainly spent fixing a performance regression in the regex
> engine on some long strings, whereby anchored patterns (e.g. ones starting
> /^.../ or /\G.../) and which contained both a fixed and floating substring
> (e.g. /^\d{4}fixed.*floating/) would sometimes prefer to check whole of the
> long string for the presence of "floating" rather than do a quick check
> and reject if "fixed" is isn't at the expected offset.
>
> I also did some tweaking of the regex engine's super-liner cache code.
>
> 1:00 GH #24258 dist/threads/t/free.t: Rare test failure in debugging build on FreeBSD
> 19:57 GH #24272 PDF parsing task running for hours instead of minutes since Perl 5.20
> 1:22 GH #24467 ExtUtils-ParseXS-3.61+: CPP directives after MODULE in same paragraph mishandled
> 6:55 make regex super-linear-cache 64-bit
> ------
> 29:14 TOTAL (HH::MM)
https://news.perlfoundation.org/post/maintaining_perl_dave_mitchell_june_2026
Thanks Dave, +1 from me.
-- Matthew Horsfall (alh)
--eda9bc239f817ba1db5a58e46d12bafcbfaa6ddd
Content-Type: text/html
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE html><html><head><title></title></head><body><div><br></div><d=
iv><br></div><div>On Tue, Jul 14, 2026, at 7:20 AM, Dave Mitchell wrote:=
</div><blockquote type=3D"cite" id=3D"qt" style=3D""><div>Last month was=
mainly spent fixing a performance regression in the regex</div><div>eng=
ine on some long strings, whereby anchored patterns (e.g. ones starting<=
/div><div>/^.../ or /\G.../) and which contained both a fixed and floati=
ng substring</div><div>(e.g. /^\d{4}fixed.*floating/) would sometimes pr=
efer to check whole of the</div><div>long string for the presence of "fl=
oating" rather than do a quick check</div><div>and reject if "fixed" is =
isn't at the expected offset.</div><div><br></div><div>I also did some t=
weaking of the regex engine's super-liner cache code.</div><div><br></di=
v><div> 1:00 GH #24258 dist/threads/t/free=
.t: Rare test failure in debugging build on FreeBSD</div><div> &nbs=
p; 19:57 GH #24272 PDF parsing task running for hours instea=
d of minutes since Perl 5.20</div><div> 1:=
22 GH #24467 ExtUtils-ParseXS-3.61+: CPP directives after MODULE in same=
paragraph mishandled</div><div> 6:55 make=
regex super-linear-cache 64-bit</div><div> ------</di=
v><div> 29:14 TOTAL (HH::MM)</div></blockquote><=
div><br></div><div><a href=3D"https://news.perlfoundation.org/post/maint=
aining_perl_dave_mitchell_june_2026">https://news.perlfoundation.org/pos=
t/maintaining_perl_dave_mitchell_june_2026</a></div><div><br></div><div>=
Thanks Dave, +1 from me.</div><div><br></div><div>-- Matthew Horsfall (a=
lh)</div></body></html>
--eda9bc239f817ba1db5a58e46d12bafcbfaa6ddd--