In perl.git, the branch smoke-me/khw-smoke has been created

<http://perl5.git.perl.org/perl.git/commitdiff/18dafef10ac4945dd99619dc03e3d394f5cbba18?hp=0000000000000000000000000000000000000000>

        at  18dafef10ac4945dd99619dc03e3d394f5cbba18 (commit)

- Log -----------------------------------------------------------------
commit 18dafef10ac4945dd99619dc03e3d394f5cbba18
Author: Karl Williamson <[email protected]>
Date:   Fri Nov 25 08:35:30 2016 -0700

    XXX -m32 fix warnings

M       utf8.c

commit c99a07435c605a56719cfbf5d5b639ad7fb191db
Author: Karl Williamson <[email protected]>
Date:   Sat Dec 10 20:24:15 2016 -0700

    embed.fnc: Make some functions not flagged as 'A'
    
    As explained in the previous commit, I misunderstood the available scope
    of functions not otherwise qualified by a flag.  This changes some of
    them to correspond with my new understanding.

M       embed.fnc
M       embed.h

commit 22069e1fac27312fd4b2532be9fa12e2f9e6c42a
Author: Karl Williamson <[email protected]>
Date:   Sat Dec 10 20:19:39 2016 -0700

    embed.fnc: Add comments; wordsmith others
    
    After some years of working on the Perl core, I did not realize that the
    default for functions in this file is to make them invisible outside the
    core.  I've jumped through hoops to try to do that, not realizing it was
    automatic.  This adds comments so newbies don't make the same mistake.
    
    And it rewords slightly various other comments.

M       embed.fnc

commit 88dbc1b82eb4f7d77e3e62881eb0aa0855f0b138
Author: Karl Williamson <[email protected]>
Date:   Sat Dec 10 19:16:56 2016 -0700

    utfebcdic.h: Follow up to adding const qualifiers
    
    This is a follow-up to commit 9f2eed981068e7abbcc52267863529bc59e9c8c0,
    which manually added const qualifiers to some generated code in order to
    avoid some compiler warnings.  The code changed by the other commit had
    been hand-edited after being generated to add branch prediction, which
    would be too hard to program in at this time, so the const additions
    also had to be hand-edited in.
    
    The commit just before this current one changed the generator to add the
    const, and I then did comparisons by hand to make sure the only
    differences were the branch predictions.  In doing so, I found one
    missing const, plus a bunch of differences in the generated code for
    EBCDIC 037.  We do not currently have a smoker for that system, so the
    differences could be as a result of a previous error, or they could be
    the result of the added 'const' causing the macro generator to split
    things differently.  It splits in order to avoid size limits in some
    preprocessors, and the extra 'const' tokens could have caused it to make
    its splits differently.
    
    Since we don't have any smokers for this, and no known actual systems
    running it, I decided not to bother to hand-edit the output to add
    branch prediction.

M       utfebcdic.h

commit e13405c1232e9d5e27fd59a2cdb7227df7f1b9d3
Author: Karl Williamson <[email protected]>
Date:   Sat Dec 10 19:09:19 2016 -0700

    regen/regcharclass.pl: Add const cast
    
    This is a follow-up to commit 9f2eed981068e7abbcc52267863529bc59e9c8c0,
    which manually added const qualifiers to some generated code in order to
    avoid some compiler warnings.  This changes the code generator to use
    the same 'const' qualifier generally.  The code changed by the other
    commit had been hand-edited after being generated to add branch
    prediction, which would be too hard to program in at this time, so the
    const additions also had to be hand-edited in.

M       regcharclass.h
M       regen/regcharclass.pl

commit a9c03686619492d9ba2f8569b1b673aad8fcac21
Author: Karl Williamson <[email protected]>
Date:   Thu Dec 8 22:01:16 2016 -0700

    PATCH: [perl #37836] Simple Regex causes SEGV
    
    As mentioned in the notes in the ticket, the test for this was wrong,
    relying on an optimization in the regex compiler that avoided the
    problematic code altogether.
    
    When corrected, the test would not segfault, but does croak.  This is
    because the regex engine croaks on malformed UTF-8, which this is.
    croaking is probably the best option here, and has been in place for a
    number of releases.  To fix the test, the pattern is modified by adding
    a character to circumvent the optimization, and the croak is guarded by
    a fresh_perl

M       t/re/pat_rt_report.t

commit 98128067be46c2aafd3c4fd86af76f92dab9d546
Author: Karl Williamson <[email protected]>
Date:   Tue Dec 6 10:16:10 2016 -0700

    toke.c: Use strpbrk instead of sequential strchr()
    
        strpbrk("ABC")
    
    does the same thing (faster) as
    
        strchr('A') || strchr('B') || strchr('C')

M       toke.c

commit 7c36b3b1cde26c98cc1633ceb31ea913e6d3ba4c
Author: Karl Williamson <[email protected]>
Date:   Fri Dec 9 13:47:27 2016 -0700

    regexec.c: Remove obsolete expression
    
    The flags here are obsolete, as UTF8_ALLOW_DEFAULT is what we want.
    It allows no more than the and'ed UTF8_ALLOW_ANYUV.

M       regexec.c

commit f812104cead2c81fd4f9d410896f2c2791153363
Author: Karl Williamson <[email protected]>
Date:   Fri Dec 9 13:25:05 2016 -0700

    regexec.c: Remove use of obsolete flag
    
    This flag is now defined as 0, so it does nothing.  This is the final
    use of it in core.

M       regexec.c
-----------------------------------------------------------------------

--
Perl5 Master Repository

Reply via email to