In perl.git, the branch maint-votes has been updated <http://perl5.git.perl.org/perl.git/commitdiff/9bb63f2b0120b9934ed08f4b4ecf689c7d1762ed?hp=6bea87da8bf8395bd2bcc06ae0c7844e7bf91ea5>
- Log ----------------------------------------------------------------- commit 9bb63f2b0120b9934ed08f4b4ecf689c7d1762ed Author: Steve Hay <steve.m....@googlemail.com> Date: Tue Aug 8 13:58:55 2017 +0100 Cross off the Time-HiRes commits: I've now pulled in 1.9741 M votes-5.24.xml commit 74a979551283ef35dcb580fad10568bce29beaa9 Author: Steve Hay <steve.m....@googlemail.com> Date: Tue Aug 8 13:52:30 2017 +0100 Vote for a test fix With that fix in as well, and with a correction to the auto-merge of bb152a4b44 (note now added about it to remind me for later), all these commits cherry-pick cleanly and everything builds and passes all tests for me. Just need some more votes now :-) M votes-5.26.xml ----------------------------------------------------------------------- Summary of changes: votes-5.24.xml | 14 -------------- votes-5.26.xml | 5 ++++- 2 files changed, 4 insertions(+), 15 deletions(-) diff --git a/votes-5.24.xml b/votes-5.24.xml index 840fec5412..a6031773a5 100644 --- a/votes-5.24.xml +++ b/votes-5.24.xml @@ -63,15 +63,6 @@ There doesn't seem to be a blead commit for this ticket, so nothing to backport. <commit votes="khw arc" id="753b2c6a60a81dacbe59e2041e30e8302484dc2d" desc="Fix -Dr output regression"/> -NOTE from jhi: so the Time-HiRes in 5.24.1 has the broken version of Time-HiRes that is broken in OS X. Possibilities: -[1] The smallish fix from https://rt.perl.org/Ticket/Display.html?id=128445 (blead c4a535af41823cc84534bf8dbecfdcc3eb7daa86) which upgrades 1.9734 to 1.9739 -[2] Craig Berry's fix in https://marc.info/?l=perl5-porters;m=148305290013791 which upgrades 5.24 maint to version 1.9741, which has now been out for two months, and seems to be doing well: http://ma ... [44 chars truncated] -NOTE from arc: I'd like to see this fixed in maint-5.24. Since the changes -in 1.9739 are already fairly involved, I'm inclined to think that the -additional risk of going to 1.9741 is limited. I therefore tentatively cast -a vote in favour of that approach. -<commit votes="jhi craigb xsawyerx alh" id="" desc="[cpan #120032] Doesn't compile with Perl 5.24.1 on OS/X"/> - NOTE from jhi: the following cluster of hexfp-related fixes (in ascending time order) only makes sense as a unit, they all touch the same code. I tried cherry-picking these onto maint-5.24, and only @@ -110,11 +101,6 @@ so I've voted for the cluster in total. <div><h3>Build and Installation Fixes</h3> -Should we not just backport 2d41a263 instead? And does 5.22 need this, too? --sprout -The entry above about [cpan #120032] is the same underlying issue, I think. --arc -See also https://rt.perl.org/Ticket/Display.html?id=128972, on which it is said that 5.22 doesn't suffer the problem (assuming it is also the same thing). -<commit votes="" id="c4a535af41823cc84534bf8dbecfdcc3eb7daa86" desc="[perl #128972] Build failure on OS X 10.12 Sierra."/> - <commit votes="craigb jhi" id="f6a154ae766a3404d83b81448ca6a356d30198e1" desc="Treat VSI C the same as DEC/Compaq/HP C."/> </div> diff --git a/votes-5.26.xml b/votes-5.26.xml index 5d2bbde165..a76a73414d 100644 --- a/votes-5.26.xml +++ b/votes-5.26.xml @@ -34,13 +34,16 @@ The same criteria apply to code in dual-life modules as to core code.) <commit votes="khw steveh" id="1d5030e143202c1e963e1fc91eb6f3afaa2df83e" ticket="131646" desc="[perl #131646] 5.26 regression: Assert fail instead of proper error msg" /> Fixes a problem with the previous patch: -<commit votes="tonyc" id="9c6b56dc65cdd9256fbe04a7baf4f085db1c04dd" ticket="131646" desc="(perl #131646) make the test less fragile" /> +<commit votes="tonyc steveh" id="9c6b56dc65cdd9256fbe04a7baf4f085db1c04dd" ticket="131646" desc="(perl #131646) make the test less fragile" /> <commit votes="steveh" id="790acddeaa0d2c73524596048b129561225cf100" ticket="131085" desc="[perl #131085] Crash with sub-in-stash" /> <commit votes="steveh" id="d7e3f70f30811328d2f6ae57e5892deccf64d0b2" ticket="105930" desc="set up catchable runloops early enough" /> <commit votes="steveh" id="655f5b268af8bf50c44ba4ae4803a33c9b792b8b" ticket="130861" desc="S_require_tie_mod(): use a new stack" /> <commit votes="steveh" id="42752acc4959c5b770bbc29532bf2677f4533c4e" ticket="128207" desc="Make LOCK_LC_NUMERIC_STANDARD recursive" /> <commit votes="steveh" id="9604fbf0722bd97ca6031a263c50ad52b6633db7" ticket="131526" desc="(perl #131526) don't go beyond the end of the NUL in my_atof2" /> + +Take care when cherry-picking this: The test count isn't adjusted correctly by Git because maint-5.26 doesn't have 90c172854f (which reduced the count by one): <commit votes="steveh" id="bb152a4b442f7718fd37d32cc558be675e8ae1ae" ticket="131575" desc="don't call Perl_fbm_instr() with negative length" /> + <commit votes="steveh" id="e13dc8886fcabf88a521e8e73c358157b1fa4c8a" ticket="131627" desc="[perl #131627] Fix multideref for $x{qw/a/->$*}" /> <commit votes="steveh" id="e26c6904d9f9f5ea818e590331b14038279332d1" ticket="131645" desc="[perl #131645] Fix assert fail in pp_sselect" /> <commit votes="steveh" id="67c3640a57440a4e9e224e9164ac9f39bdc9376f" ticket="131570" desc="(perl #131570) don't skip the temps stack entry we just allocated" /> -- Perl5 Master Repository