On 5/13/2012 4:06 AM, Sisyphus wrote:
----- Original Message ----- From: "chm" <[email protected]>
To: <[email protected]>
Sent: Saturday, May 12, 2012 11:24 PM
Subject: [Perldl] CHM/PDL-2.4.10_004.tar.gz released to CPAN
This is basically the release candidate for PDL to
support the perl 5.16 release coming next week.
Please report any problems as soon as possible
and test with the 5.16 release candidate if at all
possible.
Only problems I've found are:
#####################################
The proj-4.8.0 library breaks the proj stuff in PDL. (This is for all
perl versions - not just 5.16.)
Afaik:
1) There's still no portable fix for this;
2) Currently, the only portable solution is to work around the problem
by manually copying the 4.8.0 version of projects.h from the source to
the location into which 'make install' placed proj_api.h. (I guess an
equivalent workaround is to hack the 4.8.0 source so that 'make install'
installs projects.h.)
#####################################
On some of my builds of perl (including 32-bit 5.16) I get the following
error from inline-comment-test.t:
t/inline-comment-test.t ..... 1/3 # Failed test 'WART: commenting out a
thread
loop does not work'
# at t/inline-comment-test.t line 56.
# $a is
# [
# [0 1 2]
# [3 4 5]
# [6 7 8]
# ]
# and $b is
# [
# [1 2 3]
# [4 5 6]
# [7 8 9]
# ]
# Looks like you failed 1 test of 3.
t/inline-comment-test.t ..... Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/3 subtests
No problem with some of my other perls, however. (Haven't tried to
investigate as yet.)
The failure actually causes 'dmake test' to fail. Is that as intended ?
(I'm wondering if the tests are supposed to be conducted under 'TODO'
... in which case a test failure should not cause 'dmake test' to fail.)
I think that is the plan. I'll put that on the
list for the 2.4.11 release.
#####################################
With t/flexraw.t (all perls, I think) I'm getting warnings about there
being no SIGBUS on Windows, so I've pushed a change to FlexRaw.pm to fix
that.
In so doing, I don't *think* I've changed the scope of the 'local' in
'local $SIG{BUS}' (in FlexRaw.pm), but others might want to check that.
Thanks for fixing this. It was a side effect of
the new mmap support. win32 doesn't have all the
signals of unix and hence the warning.
The same test script also produces the following warnings:
set_data_by_file_map: calling
sys_map(SCALAR(0x2ff3dd4),48,3,0,GLOB(0x2ff3ce4),0)
set_data_by_file_map: length ${$pdl_dataref} is 48.
and
set_data_by_file_map: calling
sys_map(SCALAR(0x2fed64c),24,3,0,GLOB(0x2ff3d54),0)
set_data_by_file_map: length ${$pdl_dataref} is 24.
I haven't yet done anything about that. I couldn't immediately see where
those warnings are coming from. Do we want to eradicate them ? (I guess
so.)
Those are some diagnostic prints for the new mmap
support. We'll just need to change script to set
$PDL::debug to 0.
#####################################
Finally, perl-5.16 has deprecated 'defined(@array)'. I've pushed a
change to PGPLOT/Window.pm to eliminate the warning that deprecation has
triggered.
Cheers,
Rob
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 10.0.1424 / Virus Database: 2425/4995 - Release Date: 05/12/12
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
