In perl.git, the branch smoke-me/davem/cpan-build2 has been created
<http://perl5.git.perl.org/perl.git/commitdiff/c8594ce5356dfb36fc83c346ea2c96ff379af54e?hp=0000000000000000000000000000000000000000>
at c8594ce5356dfb36fc83c346ea2c96ff379af54e (commit)
- Log -----------------------------------------------------------------
commit c8594ce5356dfb36fc83c346ea2c96ff379af54e
Author: David Mitchell <[email protected]>
Date: Fri Dec 12 18:28:57 2014 +0000
Unicode::Collate: reduce noise in STDERR
output a note on STDOUT rather than STDERR.
M cpan/Unicode-Collate/Makefile.PL
commit ea58445510356ab0d11f8e7626bc4b19fa5e7faa
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 17:46:18 2014 +0000
Time::Piece: silence some compiler warnings
There were a few "signed > unsigned" comparisons, and a couple
of static identifier strings weren't used. Silencing the latter
was problematic for achieving it on both gcc and clang' in the end
I created a fake function that uses the unused vars.
M cpan/Time-Piece/Piece.xs
commit a97c901b040d7ee31db81778171730cd47e0f1d5
Author: David Mitchell <[email protected]>
Date: Tue Dec 16 12:53:19 2014 +0000
Time::HiRes: fix compiler warnings
On some platforms the 1st arg to get/setitimer is an enum, which
causes -Wc++-compat to complain about passing an int instead.
Since we don't portably know the enum type (on my linux system
it's __itimer_which_t), we cant reliably cast it. So just temporarily
disable -Wc++-compat.
M cpan/Time-HiRes/HiRes.xs
commit 80a276052b2f74fe8ea97be487acf7e88df4ac30
Author: David Mitchell <[email protected]>
Date: Fri Dec 12 16:30:49 2014 +0000
Test::Simple: silence some STDERR noise
t/Behavior/MonkeyPatching_diag.t calls diag() a few times, which causes
non-error stuff to appear on STDERR. I've silenced this by using
Test::Stream->intercept_start/stop, but I'm not at all familiar with the
IO model in Test::Simple and so don't know whether this is the best way to
do it
Ditto t/Test-Stream-Event-Diag.t.
M cpan/Test-Simple/t/Behavior/MonkeyPatching_diag.t
M cpan/Test-Simple/t/Test-Stream-Event-Diag.t
commit eea4f22e8fd74d7ab262d816573d71c2f4028387
Author: David Mitchell <[email protected]>
Date: Tue Dec 16 13:55:13 2014 +0000
Test::Simple: avoid warnings in subtest tests
Test::Simple can produce warnings when all subtests are skipped, e.g.
SKIP_ALL in subtest could not find flow-control label, using exception
for flow control
SKIP_ALL in subtest via 'BEGIN' or 'use', using exception for flow
control
(whether these should be warnings rather than going to STDOUT, I don't
have an opinion on).
A couple of tests in Test::Simple exercise this, resulting in output
to STDERR. Instead, trap the warnings and test that they are what we
expect.
M cpan/Test-Simple/t/Behavior/skip_all_in_subtest.t
M cpan/Test-Simple/t/Legacy/subtest/basic.t
commit f610c8f52e5d2a1dd938918f01770b8c7ce675c1
Author: David Mitchell <[email protected]>
Date: Tue Dec 9 10:30:47 2014 +0000
Test::Simple: less test noise on STDERR
Output info about successful forking to STDOUT rather than STDERR, which
should be reserved for when things have gone wrong.
M cpan/Test-Simple/t/Test-Stream-ForceExit.t
commit 43236812759391dd6e929966b8360cee6695c21d
Author: David Mitchell <[email protected]>
Date: Tue Dec 16 12:58:15 2014 +0000
Test::Harness: don't pollute STDERR
Make the informational message
No saved state, selection will be empty
appear on stdout rather than stderr.
M cpan/Test-Harness/lib/App/Prove/State.pm
commit 2a3b78d371988ce2ad904f19c0c922d04fe418e5
Author: David Mitchell <[email protected]>
Date: Sat Dec 6 14:17:59 2014 +0000
Pod::Perldoc: don't make perldoc.1 during build
During the build of perl core, cpan/Pod-Perldoc tries to make perldoc.1,
but since lots of modules haven't been built yet, MM.pm fails to run
Pod::Man and spews warnings to STDERR. So skip it if PERL_CORE is true.
M cpan/Pod-Perldoc/Makefile.PL
commit dcbbd65611affd303aa94517f489c00f8ce85378
Author: David Mitchell <[email protected]>
Date: Tue Dec 9 10:46:25 2014 +0000
Module::Metadata: less test noise on STDERR
Output info about tmp dir creation to stdout rather than stderr, which
should be reserved for when things have gone wrong.
M cpan/Module-Metadata/t/metadata.t
commit 1af4b4dcf982990d0fd23f2311adfe3217762842
Author: David Mitchell <[email protected]>
Date: Wed Dec 17 14:33:07 2014 +0000
Scalar-List-Utils: avoid warning
On clang with Address Sanitizer, ref addresses can be > 2**31,
so avoid these warnings:
Hexadecimal number > 0xffffffff non-portable
The code already did 'local $^W' but apparently this isn't enough.
M cpan/Scalar-List-Utils/t/refaddr.t
commit da444a5f42f0259109c2e6c74734ba940202d9df
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 17:19:06 2014 +0000
List::Util: silence some compiler warnings
* the dMULTICALL stuff defines and sets a variable newsp which
isn't subsequently used
* in a couple of places the vars a and b are declared and set to
GvSV(agv) etc, but not subsequently used; get rid of them;
* in one place, the compiler is too dumb to determine that in all cases,
retval is initialised; so manually initialise it at the start.
M cpan/Scalar-List-Utils/ListUtil.xs
commit 2fbe12e5cc328e0cec4e54fd531b2afaf0cea5fe
Author: David Mitchell <[email protected]>
Date: Mon Dec 8 12:43:04 2014 +0000
libnet: less test noise on STDERR
Change the tests so that information about the port a server has started
on etc, goes to to STDOUT rather than STDERR - which should be reserved for
when things have gone wrong.
M cpan/libnet/t/nntp_ipv6.t
M cpan/libnet/t/nntp_ssl.t
M cpan/libnet/t/pop3_ipv6.t
M cpan/libnet/t/pop3_ssl.t
M cpan/libnet/t/smtp_ipv6.t
M cpan/libnet/t/smtp_ssl.t
commit be94258ef43661dd0a03442818e0882867d00979
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 16:47:30 2014 +0000
Filter::Util::Call: silence two compiler warnings
There was a (cond ? unsigned : signed); and the variable
'package' is set, but not used, in unimport().
M cpan/Filter-Util-Call/Call.xs
commit 59148d85ddfdcf59d346046089170696d6448285
Author: David Mitchell <[email protected]>
Date: Tue Dec 9 11:08:49 2014 +0000
EU::MakeMaker: reduce clutter on STDERR
Don't output to STDERR the message 'Processing hints file foo',
as STDERR should be reserved for warnings and errors.
Instead, sent it to STDOUT - and even then, only with $Verbose,
as otherwise it appears as spurious output when running t/hints.t
M cpan/ExtUtils-MakeMaker/lib/ExtUtils/MakeMaker.pm
M cpan/ExtUtils-MakeMaker/t/hints.t
commit 0706f4eb9089b25e74e789eb2bbaaec8ae36235b
Author: David Mitchell <[email protected]>
Date: Wed Dec 17 13:49:19 2014 +0000
ExtUtils::Install: avoid find warnings in -j build
Avoid warnings like:
Can't cd to (/foo/cpan/ExtUtils-Install/../../lib/auto/) ExtTest:
No such file or directory
at /foo/cpan/ExtUtils-Install/../../lib/ExtUtils/Installed.pm line
201.
on parallel 'make tests' when running a parallel build in perl core.
M cpan/ExtUtils-Install/lib/ExtUtils/Installed.pm
commit 07d5956359d57f7423e7b2897a99ffc457a93b0c
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 16:14:13 2014 +0000
EU::Constant: avoid 'uninit' warning
The code generated by ExtUtils::Constant can look something like:
static int
constant (..., IV *iv_return) {
switch (...) {
case ...:
*iv_return = ...;
return PERL_constant_ISIV;
...
}
}
{
int type;
IV iv;
type = constant(..., &iv);
switch (type) {
case PERL_constant_ISIV:
PUSHi(iv);
...
}
}
and the compiler isn't clever enough to realise that the value of iv
is only used in the code path where its been set.
So initialise it to zero to shut gcc up. Ditto nv and pv.
M cpan/ExtUtils-Constant/lib/ExtUtils/Constant.pm
commit fefa09fd080dd3980f565dc40b38e5e53f4ed65b
Author: David Mitchell <[email protected]>
Date: Tue Dec 16 15:31:45 2014 +0000
encoding::warnings: don't warn about ${^ENCODING}
Assigning a non-undef value to ${^ENCODING} is deprecated, but presumably
encoding::warnings is allowed to do this.
M cpan/encoding-warnings/lib/encoding/warnings.pm
commit 807b8dc936b34de3b7757e13d2e9fdfd8ce1a0bb
Author: David Mitchell <[email protected]>
Date: Tue Dec 16 17:22:13 2014 +0000
Encode::Unicode: avoid 'always true' warning
Introduce an attr_true() variant of the attr() macro to replace uses of
SvTRUE(attr(...)).
This avoids the code trying to determine the truthiness of &PL_sv_undef,
with the associated 'always true' warning.
M cpan/Encode/Unicode/Unicode.xs
commit b274c9a468d036ebf33eccc889d0e55b6a1c617e
Author: David Mitchell <[email protected]>
Date: Fri Dec 5 16:28:33 2014 +0000
Encode/bin/enc2xs: don't write to STDERR
Progress reports aren't errors or warnings, so they should
got to STDOUT instead.
M cpan/Encode/bin/enc2xs
commit 5d96efc622b7788d534626eb64b1e2f6568bd509
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 13:38:09 2014 +0000
update t/porting/customized.dat for MD5 mod
M t/porting/customized.dat
commit f378f39d949aa3966ca4e7e155ad0acdb11be9e2
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 13:04:49 2014 +0000
Digest::MD5: silence some compiler warnings
In
SvOBJECT(sv) && strEQ(HvNAME(SvSTASH(sv)), "Digest::MD5")
The compiler isn't psychic enough to know that HvNAME() will be non-null,
so assign it to a temp var and test for nullness.
M cpan/Digest-MD5/MD5.xs
M cpan/Digest-MD5/t/files.t
commit 03083504963f6a91bd48e3b362f1deea4de1622b
Author: David Mitchell <[email protected]>
Date: Fri Dec 12 17:01:07 2014 +0000
Devel::PPPort: silence some build warnings
Some entries of the form
=for apidoc PL_parser-E<gt>linestr
have crept into the perl source (documenting structs rather than functions
or macros). Make PPPort silently ignore them rather than warning.
Whether it's wise for there to be such apidoc entries, and whether
its wise for PPPort to ignore them, I don't know.
M cpan/Devel-PPPort/parts/ppptools.pl
commit bd92714c85d7b1ddcc8ba81b85ff12df9256f254
Author: David Mitchell <[email protected]>
Date: Mon Dec 15 16:59:49 2014 +0000
Compress::Raw::Zlib: silence some compiler warnings
Some function args aren't used sometimes.
M cpan/Compress-Raw-Zlib/Zlib.xs
-----------------------------------------------------------------------
--
Perl5 Master Repository
