develooper Front page | perl.perl5.porters | Postings from September 2013

TONYC TPF grant report #10

From:
Tony Cook
Date:
September 16, 2013 11:39
Subject:
TONYC TPF grant report #10
Message ID:
20130916113920.GC14967@mars.tony.develop-help.com
[Hours]         [Activity]
2013/08/26      Monday
 0.95           #119413 - apply proposed change to blead, perldelta
 1.72           #117265 - review, minor cleanups, rebase, retest
 0.55           #117265 - apply and perldelta
 0.22           #117355 - review discussion
 0.17           #119355 - review discussion
 1.65           #85104 - fix skip handling, test, apply, perldelta
=====
 5.26

2013/08/27      Tuesday
 0.50           #119413 - respond to comment
 0.93           scan for older tickets with/without patches to 74864
 1.37           #116264 - review, test, comment
 0.22           #119351 - fix the test note some more
 1.00           #81586 - test on various perls, build blead to test,
                comment
 1.08           #118059 - test theory, try to figure out fp dupping
=====
 5.10

2013/08/28      Wednesday
 3.47           #118059 - more testing, B::SPECIAL discussion
 1.50           #118059 - test with debug tools, but cannot reproduce with
                debugging tools
=====
 4.97

2013/08/29      Thursday
 2.05           #118059 - test osfhandle results, comment
 0.85           #115928 - read metaconfig docs
 1.98           #115928 - manually edit Configure etc, testing
=====
 4.88

Which I calculate is 20.21 hours.

Approximately 10 tickets were worked on, and 3 patches were
applied.

This week I finally applied my variant of Reini's safesyscalls patch
from [perl #117265].  This change makes calls to open() and other
system calls fail if the path contains an embedded NUL ("\0")
character, and optionally warns.

I spent some more time trying to work out the race condition in socket
creation and destruction on Win32 for [perl #118059].  This script
fails rarely for me, which makes it difficult to test, and has never
failed for me when I add extra debugging output before the accept() in
the parent thread or between the connect() and exit() in the child
thread, which makes it fun to debug.  My next approach will probably
be attempting to reproduce the problem with threads rather than
Win32's pseudo-fork.



nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About