Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Commits on Feb 20, 2014
  1. C 1.45_06 my_perl_run: enable my_leave_scope, tested 5.8.4-5.16

    authored
    Filter out all free calls to static SVs/OPs on the final stack cleanup.
    
    We really only need my_leave_scope at fast_perl_destruct and at
    the end of my_perl_run to protect cv_undef with t/testc.sh 169 #303
    (Attribute::Handler leaving CVs on the stask).
    
    Use old 5.14 leave_scope code, not the new ARG[012] macros
    we use a push/pop, not direct access. and we do want to reliably
    
    Fixes #303, just not with 5.18 (static padlist) and 5.10.1 threaded yet.
Commits on Feb 19, 2014
  1. dos2unix t/z_pod.t

    authored
  2. t/manifest.t: added (unix only)

    authored
    compare MANIFEST to `git ls-tree -r --name-only HEAD >MANIFEST.git`
  3. C: revert 205fbbc Special-case io_close, move PerlIO_cleanup after sv…

    authored
    …_clean_objs
    
    the cause for the io_close problem was that PerlIO_cleanup was called before sv_clean_objs.
    global destruction -O3 looks better now
    Fixes #302 CORE--op--utf8cache.t, CORE--op--lex_assign.t, CORE--io--openpid.t
  4. C: simplify dynamic lexwarns - savepvn

    authored
    use a single line, even the old Copy method had an ending \0.
    t/modules.t: TODO Attribute::Handlers on 5.18 and 5.10 threaded #169
    all coretests pass
  5. C 1.43_05: Special-case io_close() on -O3 or B::CC for 5.8 (#303)

    authored
    add my_io_close to @static_free on 5.8 and blessed filehandles
    to close the static file or dir before sv_clean_objs() tries to free the static xpvio.
    Fixes testcc 22
  6. untodo coretests

    authored
    3 new failing SIG-302 tests,
    passing: t/C-COMPILED/CORE--op--gv.t
  7. 5.18 bytecode todos

    authored
  8. t/issue212.t: test 3 fixed by 1.45_04 b1e949a

    authored
    non-destructive get_isa for 5.8
  9. t/issue197.t: adjust test 4 TODO

    authored
    no special cases 5.16+5.14 which pass 2 but fail 4
Commits on Feb 18, 2014
  1. log.modules-5.006002-nt 1.45_04 c61065d

    authored
    no changes: 16 fails
  2. log.modules-5.006002-nt 1.45_04 b1e949a

    authored
    worse
  3. C: fix use warnings (was missing from lexwarn optim)

    authored
    simplify special CvSTART cases: start cop is always stored, with more local info
  4. adjust TODOs: 274, 59

    authored
Commits on Feb 17, 2014
  1. C 1.43_04: non-destructive get_isa for 5.8

    authored
    fixes #210 on 5.8
    dont create and store an empty ISA when it does not exist
  2. C: save IO for *STDIN also

    authored
    mainly the utf8 layer (not yet working)
    add testc.sh 210
  3. module updates: 1.45_03 79fa3ed

    authored
    dramatic regressions >5.12, but much better with 5.8
    no proper env for 5.6 on this smoker
  4. Changes: headline

    authored
  5. CC: fix test 48 for 5.18 threaded CopSTASHPV_set NULL

    authored
    use now CopSTASHPV_set, and add 5.16 special case from B::C
Commits on Feb 15, 2014
Commits on Feb 14, 2014
  1. C: void perl_destruct on 5.6

    authored
  2. t/C-COMPILED/template.pl: revert optim specific bin names

    authored
    we need to check in the tests for $0 for comparing diagnostics
  3. C: fix for empty static lexwarn<n>

    authored
    e.g. t/CORE/io/utf8.t
  4. fast-testing: add --watch option

    authored
    it disturbs other tests
  5. t/fast-testing.sh: fixed for bash-3

    authored
    cproc or not decided automatically
  6. C: cop file, stash and lexwarns are now all dynamic

    authored
    get rid of static inits,
    static_free cop_list is now dead code
    protect from double initialization for a CvSTART cop or CvROOT cop
  7. C: 1c5062f accidently enabled -ffast-destruct on -O0

    authored
    fix TODOs for t/issue208.t
    -O3 still broken > 5.13
  8. C 1.45_03: Keep internal packages if used #300

    authored
    if savecv detects an internal core function, keep the package
    and prevent it from being optimzed away, i.e. mro with test #300.
    check %include_package
Something went wrong with that request. Please try again.