develooper Front page | perl.perl5.porters | Postings from June 2014

TONYC TPF grant report #40

Tony Cook
June 24, 2014 11:13
TONYC TPF grant report #40
Message ID:
[Hours]         [Activity]
2014/06/09      Monday
 0.55           #122046 review patch, discussion, apply to blead
 0.58           #122034 review, test and apply to blead
 0.27           #121932 reject ticket
 0.33           #121746 review and ask jhi about CPAN ticket
 0.50           #121743 check if patch has been applied, find it has and
 0.28           #119453 double-check patch and push smoke-me
 0.63           #115932 review and try to apply
 0.35           #121474 re-review patch and discussion and reject
 1.23           #121732 try to diagnose unrelated minitest brokeness,
                testing and fixes on cygwin

2014/06/10      Tuesday
 1.08           #121733 apply, testing and fixes on cygwin, testing on
                win32/linux, push to blead
 0.30           #119453 retest and apply to blead
 1.27           regen speed discussion, cygwin compiler warning fixes
 2.70           regen profiling, optimization

2014/06/11      Wednesday
 1.48           more regen optimization, stall
 1.03           fix bugs in POSIX::strxfrm()
 1.55           #122000, #122009 find the right opebbsd VM, rewrite
                author, re-test, add author entry, push to blead
 0.22           #122002 comment
 0.70           #121505 setup prod gcc 4.9

2014/06/12      Thursday
 0.35           #122068 review, test and apply to blead
 1.28           #122070 review code, track down possible bug fix
 1.83           #122070 add test, review second patch, push to blead with
                version bumps, perldelta fixes, comment and resolve
 0.52           #121974 review and comment
 1.17           #122002 try on Solaris, debugging, comment

Which I calculate is 20.2 hours.

Approximately 16 tickets were reviewed or otherwise worked on, and 5
patches were applied.

Probably the most interesting task this week was attempting to
optimize regen/ which became noticably slower when
extra EBCDIC code pages were added to the build.  Two fairly simple
optimizations fell out of the profile from Devel::NYTProf.

get_a2n() and get_I8_2_utf() were each called over 150,000 times, and
returned their mappings as a list, while in most cases only a small
part of the mapping was used.  I changed these return an array
reference instead, speeding up significantly. Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About