In perl.git, the branch smoke-me/khw-turkish has been created
<https://perl5.git.perl.org/perl.git/commitdiff/760e2b4e29f6bcd3dcc520354ff56c5106078e1d?hp=0000000000000000000000000000000000000000>
at 760e2b4e29f6bcd3dcc520354ff56c5106078e1d (commit)
- Log -----------------------------------------------------------------
commit 760e2b4e29f6bcd3dcc520354ff56c5106078e1d
Author: Karl Williamson <k...@cpan.org>
Date: Thu Jan 31 09:28:58 2019 -0700
smoke2
commit ba3809bda5d2b4921c5ce07d950abadd5c97d9fe
Author: Karl Williamson <k...@cpan.org>
Date: Thu Jan 31 08:50:03 2019 -0700
skip
commit decddcaddeeab3c068e31363877bc48890ba6f0d
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 15:00:30 2019 -0700
regcomp.c: Fix recent optimization of [...] bug
This bug was introduced in b2296192536090829ba6d2cb367456f4e346dcc6
n 5.29.7. Using /il should not result in looking for a [:posix:] class
that matches the code points given.
commit 7ac5920d0c5639a87e5d9c8bc22e3d768d24ca9b
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 11:24:12 2019 -0700
pp.c: Don't assume worst case memory needs
Since 5.28, there has been a function that will calculate the expansion
of a string when converted into UTF-8, using per-word operations. This
means it runs 8 times faster than doing this count previously would have
taken.
I've come to believe it is better to calculate how much memory we need
than to overallocate based on worst-case scenarios. This is because in
very large strings, over allocating can lead to unnecessary inefficient
processing.
This commit changes several instances in pp.c where a string needs to be
converted to UTF-8 to not assume the worst case, but instead calculate
what's needed using the faster function.
commit 7b7dea0e808e7517055fe813005e0d361802cb41
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 11:09:01 2019 -0700
pp.c: Don't use function call for easy copy
Like the previous commit, this code is adding the UTF-8 for a Greek
character to a string. It previously used Copy, but this character is
representable as two bytes in both ASCII and EBCDIC UTF-8, the only
character sets that Perl will ever supports, so we can use the
specialized code that is used most everywhere else for two byte UTF-8
characters, avoiding the function overhead, and having to treat this
character as particularly special.
commit a51239c49c5c845428578557eb5d0d42f7ec75ec
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 10:52:41 2019 -0700
pp.c: Don't use function call for easy copy
This code is adding the UTF-8 for a Greek character to a string. It
previously used Copy, but this character is representable as two bytes
in both ASCII and EBCDIC UTF-8, the only character sets that Perl will
ever supports, so we can use the specialized code that is used most
everywhere else for two byte UTF-8 characters, avoiding the function
overhead, and having to treat this character as particularly special.
commit 9fd215dfe074eeef617bfcef12cb71419a96111d
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 10:35:21 2019 -0700
pp.c: pp_fc(): Simplify
The function being called does everything that the code being eliminated
here did. We just pass the function the final destination instead of a
temporary.
commit 8f589a86651858e5a470132f47cdb3b0d5e65285
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 10:27:17 2019 -0700
pp.c: White-space, comments only
commit 4753f7be7282ade663fbca6b0d26ae0d835614d9
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 10:02:35 2019 -0700
pp.c: Reorder && order in 'if'
This makes the test most likely to fail be first, and adding an
UNLIKELY() to it, thus saving a conditional in most instances.
commit e9d144be2a7188ca585cf6dab8301444d86a4212
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 09:56:15 2019 -0700
XXX move to /locale ws. locale.c white space only
commit 57360ede042b2efacf6481eccada1f67c8c53d23
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 30 09:08:13 2019 -0700
handy.h: Add comment
commit 062b4f9a3b1bf258a5d7cce9ce2446faa8c19a6d
Author: Karl Williamson <k...@cpan.org>
Date: Tue Jan 29 22:25:03 2019 -0700
t/re/fold_grind.pl: White-space only
Just align some logical or clauses for readability.
commit a3c6f712f764cc7d08e0f8ad70f7de9d4e2b2a6e
Author: Karl Williamson <k...@cpan.org>
Date: Tue Jan 29 22:02:59 2019 -0700
pp.c: Use faster method to convert to UTF-8
There is a special inline function that's used when converting a single
byte to UTF-8, that is faster than the more general one used prior to
this commit.
commit 4f5c56a624b055dcb6acc4f41ea8062584ddf882
Author: Karl Williamson <k...@cpan.org>
Date: Tue Jan 29 22:01:18 2019 -0700
pp.c: Add missing assert
The comments say there is an assert, but it wasn't there.
commit 4f58311d8edc99ca2efc72d16ecfd73eeed558d5
Author: Karl Williamson <k...@cpan.org>
Date: Fri Jan 25 09:55:58 2019 -0700
handy.h: White-space only
Vertically align the ternary colon with the question mark above it.
commit 73bd2eff58044ff82827af73d27236bc77673acd
Author: Karl Williamson <k...@cpan.org>
Date: Wed Jan 23 15:42:35 2019 -0700
handy.h: Add void * casts to memEQ, memNE
This change is to allow these macros to be called without having to do
casting in the call.
commit 77d933ada18a2bb08390a8af9f1291cfda8bbcb5
Author: Karl Williamson <k...@cpan.org>
Date: Mon Jan 21 09:46:00 2019 -0700
mktables: Make Turkic 'I' chars problematic
In a Turkic locale, these are problematic because their mappings
cross the 255/256 boundary.
This change has the side effect of causing U+307 to be added to the
problematic list, and it normally really isn't problematic, because in
those locales where U+130 and U+131 are problematic, U+307 isn't used.
But applications could switch in and out of Turkic locales, so it's best
to leave it be considered problematic. The consequences of making this
mark problematic are simply slightly less optimized regex pattern code.
-----------------------------------------------------------------------
--
Perl5 Master Repository
