In perl.git, the branch hv/study_chunk has been created

<http://perl5.git.perl.org/perl.git/commitdiff/ca4e68fc9a0c2fa6c95b13988cdf951ebdadce19?hp=0000000000000000000000000000000000000000>

        at  ca4e68fc9a0c2fa6c95b13988cdf951ebdadce19 (commit)

- Log -----------------------------------------------------------------
commit ca4e68fc9a0c2fa6c95b13988cdf951ebdadce19
Author: Hugo van der Sanden <[email protected]>
Date:   Tue Dec 2 14:31:55 2014 +0000

    WIP not for merging
    
    I don't understand the perf results comparing compilation of various
    patterns before and after the refactoring of study_chunk: all the
    instruction/data read/write counts appear to be identical, which
    doesn't make sense.
    
    Generated with:
      Porting/bench.pl -j=1 --raw --benchfile=t/perf/regex \
          -w=t/perf/regex-results ./blead ./study_chunk

A       t/perf/regex
A       t/perf/regex-results

commit a04ff6295074bbe8d954ae52e9588f0aa92ef695
Author: Hugo van der Sanden <[email protected]>
Date:   Sun Nov 30 11:23:05 2014 +0000

    study_chunk: extract the rest of the rck types
    
    rck_endlikish, rck_logical, rck_gpos, rck_trie, and rck_default for
    everything else

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 4353f05ced066640fda44270870eb3eb90c83fc9
Author: Hugo van der Sanden <[email protected]>
Date:   Sat Nov 29 18:18:32 2014 +0000

    study_chunk: extract rck_open, rck_close, rck_eval

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 89e37d3a7e492ca5164c2c238a03cfc657a8333c
Author: Hugo van der Sanden <[email protected]>
Date:   Sat Nov 29 16:14:15 2014 +0000

    study_chunk: extract rck_lookaround

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 58b52945f8a4dbfe4b33d8cda4f420b37bfe926c
Author: Hugo van der Sanden <[email protected]>
Date:   Fri Nov 28 01:40:34 2014 +0000

    study_chunk: extract rck_eolish

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 040854a6b70137752ecaad20a2101b57bc09dcec
Author: Hugo van der Sanden <[email protected]>
Date:   Fri Nov 28 01:26:36 2014 +0000

    study_chunk: extract rck_simple

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 7331cd59e723b5aa24f7b070714aee2ca7253c6f
Author: Hugo van der Sanden <[email protected]>
Date:   Tue Nov 25 15:36:32 2014 +0000

    study_chunk: extract rck_lnbreak

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 7872234e0c1d0562464c96946e3aa1175734903a
Author: Hugo van der Sanden <[email protected]>
Date:   Tue Nov 25 15:12:56 2014 +0000

    study_chunk: break up REGNODE_VARIES into rck_* routines

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 56f26653320dd66d1611a64aadeb835bef9ae171
Author: Hugo van der Sanden <[email protected]>
Date:   Tue Nov 25 12:04:24 2014 +0000

    study_chunk: extract rck_exactfish
    
    Also move the join_exact() call into the exact/exactfish handling.

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 2aeb791d107589b33e3a18a8e7578b854f5e31ba
Author: Hugo van der Sanden <[email protected]>
Date:   Mon Nov 24 12:30:10 2014 +0000

    study_chunk: extract rck_exact

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit c6d9b9d40025a08e92a82f0361a15249a29cbb4c
Author: Hugo van der Sanden <[email protected]>
Date:   Mon Nov 24 12:18:10 2014 +0000

    study_chunk: extract rck_suspend, rck_gosub, rck_gostart
    
    Add helper function rck_enframe; there's still common code for gosub
    and gostart that could be refactored.

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit f5569a541b4023c32335fbc5afd43869b2ca4c42
Author: Hugo van der Sanden <[email protected]>
Date:   Mon Nov 24 11:25:58 2014 +0000

    study_chunk: simplify PAREN_TEST and related macros
    
    Define PAREN_OFFSET to point to the bitvector for the relevant depth, update
    PAREN_TEST, PAREN_SET and PAREN_UNSET to take a depth instead of a pointer,
    and simplify the various users.

M       regcomp.c

commit 3472ebb728e7489371bf1a8b237d7b1955e764b8
Author: Hugo van der Sanden <[email protected]>
Date:   Sun Nov 23 16:07:52 2014 +0000

    study_chunk: split up rck_branch
    
    Give BRANCH, BRANCHJ and IFTHEN each their own peep routines; move the
    shared code to sc_peep_make_trie.

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 87277c0406d6d549fd3662c5d6852f3c6d31514b
Author: Hugo van der Sanden <[email protected]>
Date:   Sat Nov 22 14:06:23 2014 +0000

    study_chunk: extract rck_branch

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 0b0bf9dbbc5a9879758a280bdc4c69c2b3f32b2b
Author: Hugo van der Sanden <[email protected]>
Date:   Fri Nov 21 16:25:19 2014 +0000

    study_chunk: extract rck_definep

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit f27e5c264f0a52bd329747576f182a136b9aafdc
Author: Hugo van der Sanden <[email protected]>
Date:   Wed Nov 19 15:17:59 2014 +0000

    study_chunk: extract rck_elide_nothing

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit cf5123792428b8c8c02ad1cd72eb62f0bc152338
Author: Hugo van der Sanden <[email protected]>
Date:   Fri Nov 21 14:25:01 2014 +0000

    study_chunk: remove JOIN_EXACT macro
    
    It's used only once, and the code is clearer when we can see the condition
    it applies.

M       regcomp.c

commit aaf33b417b41c7e2595eed68f59d0e9f68b49d74
Author: Hugo van der Sanden <[email protected]>
Date:   Wed Nov 19 14:58:01 2014 +0000

    study_chunk: scanp is not needed internally

M       regcomp.c

commit 5065a313cc3be024a4dea3e6bd2359ff5624930a
Author: Hugo van der Sanden <[email protected]>
Date:   Wed Nov 19 14:52:25 2014 +0000

    study_chunk_one_node: reindent
    
    1900-odd lines of whitespace, deindented after removal of outer loop

M       regcomp.c

commit 5205dccf67ac5299cc0ff37faf152dac2f49fafc
Author: Hugo van der Sanden <[email protected]>
Date:   Wed Nov 19 14:41:41 2014 +0000

    study_chunk: switch study_chunk_one_frame to study_chunk_one_node
    
    Move per-frame diagnostics and outer loop up to study_chunk, leaving
    just the work to do on a single node. This requires a return value
    to show if the inner loop should be terminated.
    
    The outdent of what remains is done separately in the next commit.

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit 2f1930710a0fad72003340f1c18535705fb694b8
Author: Hugo van der Sanden <[email protected]>
Date:   Wed Nov 19 12:38:12 2014 +0000

    study_chunk: remove commented-out "exit recursion" code
    
    This was introduced in 099ec7dc then commented out in 0ac75573.

M       regcomp.c

commit 98ee55894c43ab771706ca4600382ff5d1054616
Author: Hugo van der Sanden <[email protected]>
Date:   Mon Nov 17 21:31:13 2014 +0000

    study_chunk: separate study_chunk_one_frame
    
    Removes the need for labels fake_study_recurse and finish.

M       embed.fnc
M       embed.h
M       proto.h
M       regcomp.c

commit c9e2d55003cd863bb94d51964c2d5b20634fdfe1
Author: Hugo van der Sanden <[email protected]>
Date:   Mon Nov 17 17:39:19 2014 +0000

    study_chunk: move params into struct for refactoring

M       regcomp.c
-----------------------------------------------------------------------

--
Perl5 Master Repository

Reply via email to