[libscalar-list-utils-perl] annotated tag upstream/1.44 created (now ca56675)

Salvatore Bonaccorso carnil at debian.org
Fri Mar 18 15:16:17 UTC 2016


This is an automated email from the git hooks/post-receive script.

carnil pushed a change to annotated tag upstream/1.44
in repository libscalar-list-utils-perl.

        at  ca56675   (tag)
   tagging  63a38aa1e347fc7c7ae9edc332f11e024d3caf40 (commit)
  replaces  upstream/1.43
 tagged by  Salvatore Bonaccorso
        on  Fri Mar 18 16:08:46 2016 +0100

- Log -----------------------------------------------------------------
Upstream version 1.44

Adam Herzog (2):
      Add Travis CI config.
      Keep .travis.yml out of dists.

Dagfinn Ilmari Mannsåker (1):
      Don't reference newsp on >= 5.23.8 (RT#112312)

Paul "LeoNerd" Evans (37):
      No need to include an __END__ marker; move the '1;' to end of file
      Initial hack at a uniq() function, in PP
      Move uniq() to 'OTHER FUNCTIONS' section
      Initial implementation of a uniqnum() as well
      Unit-tests for uniq() functions
      Add unit tests to check that uniq obeys stringification overloading and causes timely destruction
      Reïmplement uniq() in XS
      Test that uniqnum() can distinguish floats, not just integers
      Reïmplement uniqnum() in XS
      Test that uniqnum() correctly handles both Infs and NaN
      Have uniq*() family return count in scalar context
      Have uniqnum() reüse the same temporary number-stringification buffer rather than creating a churn of many SVt_PVs
      Optimise uniq*() for empty/1-element lists
      No need to allocate new undef SVs for seen hash values, we can just use PL_sv_undef
      Extract the 'ix' loop-invariant out of the central for loop, for performance
      Test that uniq() handles Unicode strings without warning
      Take UTF8-ness of SvPVs into account in uniq()
      Ensure correct handling in uniqnum() of UVs and IVs outside of NV range
      Ensure that uniqnum() runs magic atmost once per argument, and before it tests for Sv[UI]OK()
      Observe a KNOWN BUG that uniqnum() gets confused by oversized bignums
      Bumped version to 1.43_001; updated Changes notes
      Capture t/uniq.t in MANIFEST
      Perls before 5.14 lack SvNV_nomg()
      Don't rely on exact stringification of Inf or NaN values when testing uniqnum()
      sv_newmortal() is nicer than sv_2mortal(newSV(0))
      Apply correct indentation in t/uniq.t
      Unit test uniqnum() with numification overload
      Bumped version to 1.43_003; updated Changes notes
      Don't rely on "Inf" to generate an infinity; 1E1000 seems more reliable
      Bumped version to 1.43_004; updated Changes notes
      A (hopefully) more reliable way to generate an Inf that works even on longdouble machines
      Bumped version to 1.43_005; updated Changes notes
      Merge remote-tracking branch 'origin/master' into uniq
      Empty commit to trigger travis-ci
      Add a note about the handling of undef within uniq()
      Merge branch 'uniq'
      Bumped version to 1.44; updated Changes notes

Paul Evans (3):
      Merge pull request #39 from ilmari/newsp-fix
      Merge pull request #40 from dur-randir/uniq
      Merge pull request #37 from adamherzog/add-travis-ci

Salvatore Bonaccorso (1):
      Imported Upstream version 1.44

Sergey Aleynikov (3):
      let hv_common deal with all SV->key transformations
      use only a single hash lookup on whatever perls possible
      temporary sv for numification holds plain PV, so access it's interals directly

-----------------------------------------------------------------------

No new revisions were added by this update.

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-perl/packages/libscalar-list-utils-perl.git



More information about the Pkg-perl-cvs-commits mailing list