Re: Vanilla Perl and 5.16.0

2012-02-26 Thread kmx

Hi again,

my quick test with 5.15.8 I have mentioned in my previous post was 
probably too quick (I wrongly set up 64bit build), here are updated results:


1/ tests results for 32bit-gcc
- crash (segfault in UNIX words) in re/reg_eval_scope.t
- crash (segfault in UNIX words) in op/fork.t
- tests hang up in 
../cpan/CPANPLUS-Dist-Build/t/02_CPANPLUS-Dist-Build.t


The crashes are "new" the hang up I have already experienced with 5.14.x 
- 
http://www.nntp.perl.org/group/perl.perl5.porters/2011/04/msg171428.html 
and


2/ tests results for 64bit-gcc
- crashes (more than one) in re/reg_eval_scope.t
- crash in op/fork.t
- some strange error during op/taint.t (maybe something broken on 
my Win7 box)

- one failed test in ../ext/Pod-Html/t/cache.t

3/ pointer-to-int and int-to-pointer casting warnings with 64bit-gcc

"Only" approx 60 occurrences in the following files:
- pp_pack.c (3x)
- RealPPPort.xs (1x)
- POSIX.xs (1x)
- Socket.xs (7x)
- Storable.xs (more than 40)

4/ in order to have NDBM_File + ODBM_File in strawberry perl we need 
these two extra files in perl core tarball:

ext/NDBM_File/hints/MSWin32.pl
ext/ODBM_File/hints/MSWin32.pl
mentioned in my RT https://rt.perl.org/rt3/Ticket/Display.html?id=71680 
(please consider including them in 5.16.0)


--
kmx


Re: Vanilla Perl and 5.16.0

2012-02-26 Thread kmx

Hi Ricardo,

I have done just a quick test building 5.15.8 with gcc 4.6.2 (gcc 
candidate to be included in the next strawberry perl release, gcc 
binaries built by Mark Dootson).


The good news is that it is possible to build working perl binaries with 
both 32/64bit gcc (I am using c-runtime from mingw-w64.sf.net project).


1/ tests results for 32bit-gcc
- crash (segfault in UNIX words) in re/reg_eval_scope.t
- crash (segfault in UNIX words) in op/fork.t
- tests hang up in 
../cpan/CPANPLUS-Dist-Build/t/02_CPANPLUS-Dist-Build.t


The crashes are "new" the hang up I have already experienced with 5.14.x 
- 
http://www.nntp.perl.org/group/perl.perl5.porters/2011/04/msg171428.html 
and


2/ tests results for 64bit-gcc
- crashes (more than one) in re/reg_eval_scope.t
- crash in op/fork.t
- some strange error during op/taint.t
- but test suite finished with theese results:

Test Summary Report
---
op/fork.t   (Wstat: 
0 Tests: 25 Failed: 1)

  Failed test:  16
op/taint.t  (Wstat: 
0 Tests: 791 Failed: 2)

  Failed tests:  1, 391
../cpan/CGI/t/tmpdir.t  (Wstat: 
0 Tests: 9 Failed: 0)

  TODO passed:   3-9
../dist/Storable/t/compat01.t   (Wstat: 
1536 Tests: 6 Failed: 6)

  Failed tests:  1-6
  Non-zero exit status: 6
../dist/Storable/t/file_magic.t (Wstat: 
256 Tests: 79 Failed: 1)

  Failed test:  13
  Non-zero exit status: 1
../dist/Storable/t/malice.t (Wstat: 
512 Tests: 404 Failed: 2)

  Failed tests:  13, 132
  Non-zero exit status: 2
../ext/PerlIO-scalar/t/scalar.t (Wstat: 
65280 Tests: 76 Failed: 0)

  Non-zero exit status: 255
  Parse errors: Bad plan.  You planned 79 tests but ran 76.
../ext/Pod-Html/t/cache.t   (Wstat: 
256 Tests: 10 Failed: 1)

  Failed test:  8
  Non-zero exit status: 1
Files=2327, Tests=526521, 9710 wallclock secs (55.39 usr +  5.12 sys = 
60.51 CPU)

Result: FAIL
dmake:  Error code 141, while making 'test'

3/ pointer-to-int and int-to-pointer casting warnings with 64bit-gcc

Not sure how much you are aware of Windows (I guess that rather less 
than more) but there is "a thing" with 64 bit MS Windows that sizeof 
pointer is 8 bytes but sizeof int (or long) is just 4 bytes which might 
obviously cause some troubles.


The gcc-4.6.2 compiler I have used has by default turned on warnings:
- cast from pointer to integer of different size [-Wpointer-to-int-cast]
- cast to pointer from integer of different size [-Wint-to-pointer-cast]

During perl build on 64bit Windows there are many warnings of these kind 
(by many I mean approx 500)


--
kmx