Re: revert perl-5.26.0 update?

2017-07-30 Thread Leo Famulari 
On Fri, Jul 28, 2017 at 11:26:22PM +0200, Ludovic Courtès wrote:
> Woow, impressive patch set!  At first sight the 26 patches look
> reasonable to me, so I would suggest pushing them, along with the
> biber-next@2.7 update you mentioned.

Alright, I'll clean it up and push today.

> As for biber itself, perhaps the best option is to make biber-next the
> new biber?  I don’t use it so I can’t really tell whether there’d be
> undesirable consequences.

Ricardo added biber-2.5 in 253cdd6c1464c944be2418ee9161f5f5b57e0eee.
Ricardo, can you check if it's still necessary?

> Another option might be to mass-escape left braces in the Biber code,
> but that’s probably not that easy…

I'll try it. The upstream changes suggest it might not be that hard.

> Anyway, this patch series probably closes the main blocker for
> core-updates no?

AFAIK, but I'm only testing x86_64-linux, and I'm not running a
graphical system.


signature.asc
Description: PGP signature

Re: revert perl-5.26.0 update?

2017-07-28 Thread Ludovic Courtès 
Hi Leo,

Leo Famulari  skribis:

> On Thu, Jul 27, 2017 at 11:03:50AM +0200, Ludovic Courtès wrote:
>> Reverting is not an option at this point IMO.  There are several Date::*
>> modules required by Biber that FTBFS and need an update, indeed, but I
>> think we should rather find a way to fix them (I spent a bit of time on
>> it but then moved on to something else.)
>
> I've got the Date::* modules building with the attached patch series.

Woow, impressive patch set!  At first sight the 26 patches look
reasonable to me, so I would suggest pushing them, along with the
biber-next@2.7 update you mentioned.

As for biber itself, perhaps the best option is to make biber-next the
new biber?  I don’t use it so I can’t really tell whether there’d be
undesirable consequences.

Another option might be to mass-escape left braces in the Biber code,
but that’s probably not that easy…

Thoughts?

Anyway, this patch series probably closes the main blocker for
core-updates no?

Thanks,
Ludo’.

Re: revert perl-5.26.0 update?

2017-07-27 Thread Leo Famulari 
On Thu, Jul 27, 2017 at 06:07:09PM -0400, Leo Famulari wrote:
> On Thu, Jul 27, 2017 at 11:03:50AM +0200, Ludovic Courtès wrote:
> > Reverting is not an option at this point IMO.  There are several Date::*
> > modules required by Biber that FTBFS and need an update, indeed, but I
> > think we should rather find a way to fix them (I spent a bit of time on
> > it but then moved on to something else.)
> 
> I've got the Date::* modules building with the attached patch series.
> However, biber and biber-next fail their test suites. The last part of
> the test suite:

biber-next passes its tests when updated to 2.7, the latest upstream
version.


signature.asc
Description: PGP signature

Re: revert perl-5.26.0 update?

2017-07-27 Thread Leo Famulari 
On Thu, Jul 27, 2017 at 11:03:50AM +0200, Ludovic Courtès wrote:
> Reverting is not an option at this point IMO.  There are several Date::*
> modules required by Biber that FTBFS and need an update, indeed, but I
> think we should rather find a way to fix them (I spent a bit of time on
> it but then moved on to something else.)

I've got the Date::* modules building with the attached patch series.
However, biber and biber-next fail their test suites. The last part of
the test suite:

--
[...]
Unescaped left brace in regex is deprecated here (and will be fatal in Perl 
5.30), passed through in regex; marked by <-- HERE in m/}(\pM+\pL){ <-- HERE 
(?!\pL+\\)/ at 
/tmp/guix-build-biber-2.5.drv-0/biber-2.5/blib/lib/Biber/LaTeX/Recode.pm line 
295.
t/tool-bltxml-inout.t .. ok
Unescaped left brace in regex is deprecated here (and will be fatal in Perl 
5.30), passed through in regex; marked by <-- HERE in m/}(\pM+\pL){ <-- HERE 
(?!\pL+\\)/ at 
/tmp/guix-build-biber-2.5.drv-0/biber-2.5/blib/lib/Biber/LaTeX/Recode.pm line 
295.
t/tool-bltxml.t  ok
Unescaped left brace in regex is deprecated here (and will be fatal in Perl 
5.30), passed through in regex; marked by <-- HERE in m/}(\pM+\pL){ <-- HERE 
(?!\pL+\\)/ at 
/tmp/guix-build-biber-2.5.drv-0/biber-2.5/blib/lib/Biber/LaTeX/Recode.pm line 
295.
t/tool.t ... ok
Unescaped left brace in regex is deprecated here (and will be fatal in Perl 
5.30), passed through in regex; marked by <-- HERE in m/}(\pM+\pL){ <-- HERE 
(?!\pL+\\)/ at 
/tmp/guix-build-biber-2.5.drv-0/biber-2.5/blib/lib/Biber/LaTeX/Recode.pm line 
295.
t/uniqueness.t . ok
Unescaped left brace in regex is deprecated here (and will be fatal in Perl 
5.30), passed through in regex; marked by <-- HERE in m/}(\pM+\pL){ <-- HERE 
(?!\pL+\\)/ at 
/tmp/guix-build-biber-2.5.drv-0/biber-2.5/blib/lib/Biber/LaTeX/Recode.pm line 
295.
t/utils.t .. ok
Unescaped left brace in regex is deprecated here (and will be fatal in Perl 
5.30), passed through in regex; marked by <-- HERE in m/}(\pM+\pL){ <-- HERE 
(?!\pL+\\)/ at 
/tmp/guix-build-biber-2.5.drv-0/biber-2.5/blib/lib/Biber/LaTeX/Recode.pm line 
295.

#   Failed test 'xdata test - 1'
#   at t/xdata.t line 114.
# --- Got
# +++ Expected
# @@ -1,26 +1,26 @@
#  '\\entry{xd1}{book}{}
#\\name{author}{1}{}{%
#  {{hash=51db4bfd331cba22959ce2d224c517cd}{%
# family={Ellington},
# family_i={E\\bibinitperiod},
# given={Edward},
# given_i={E\\bibinitperiod}}}%
#}
#\\list{location}{2}{%
#  {New York}%
#  {London}%
#}
#\\list{publisher}{1}{%
#  {Macmillan}%
#}
#\\strng{namehash}{51db4bfd331cba22959ce2d224c517cd}
#\\strng{fullhash}{51db4bfd331cba22959ce2d224c517cd}
#\\field{sortinit}{E}
# -  \\field{sortinithash}{07bbd5a529b5beaa311df5be05b874bc}
# +  \\field{sortinithash}{fefc5210ef4721525b2a478df41efcd4}
#\\field{labelyear}{2007}
#\\field{datelabelsource}{}
#\\field{labelnamesource}{author}
#\\field{note}{A Note}
#\\field{year}{2007}
#  \\endentry
#  '

#   Failed test 'xdata test - 2'
#   at t/xdata.t line 115.
# --- Got
# +++ Expected
# @@ -1,29 +1,29 @@
#  '\\entry{xd2}{book}{}
#\\name{author}{1}{}{%
#  {{hash=68539e0ce4922cc4957c6cabf35e6fc8}{%
# family={Pillington},
# family_i={P\\bibinitperiod},
# given={Peter},
# given_i={P\\bibinitperiod}}}%
#}
#\\list{location}{2}{%
#  {New York}%
#  {London}%
#}
#\\list{publisher}{1}{%
#  {Routledge}%
#}
#\\strng{namehash}{68539e0ce4922cc4957c6cabf35e6fc8}
#\\strng{fullhash}{68539e0ce4922cc4957c6cabf35e6fc8}
#\\field{sortinit}{P}
# -  \\field{sortinithash}{24100cef455d7974167575052c29146e}
# +  \\field{sortinithash}{c0a4896d0e424f9ca4d7f14f2b3428e7}
#\\field{labelyear}{2003}
#\\field{datelabelsource}{}
#\\field{labelnamesource}{author}
#\\field{abstract}{An abstract}
#\\field{addendum}{Москва}
#\\field{note}{A Note}
#\\field{venue}{venue}
#\\field{year}{2003}
#  \\endentry
#  '
# Looks like you failed 2 tests of 5.
t/xdata.t .. 
Dubious, test returned 2 (wstat 512, 0x200)
Failed 2/5 subtests 

Test Summary Report
---
t/annotations.t  (Wstat: 256 Tests: 1 Failed: 1)
  Failed test:  1
  Non-zero exit status: 1
t/basic-misc.t   (Wstat: 3584 Tests: 67 Failed: 14)
  Failed tests:  1, 5, 10, 13, 16-18, 44, 57-62
  Non-zero exit status: 14
t/biblatexml.t   (Wstat: 512 Tests: 5 Failed: 2)
  Failed tests:  1, 5
  Non-zero exit status: 2
t/crossrefs.t(Wstat: 5376 Tests: 33 Failed: 21)
  Failed tests:  1-6, 8-16, 18-19, 21-22, 32-33
  Non-zero exit status: 21
t/dm-dateformats.t   (Wstat: 1792 Tests: 33 Failed: 7)
  Failed

Re: revert perl-5.26.0 update?

2017-07-27 Thread Efraim Flashner 


On July 27, 2017 9:11:57 PM GMT+03:00, Leo Famulari  wrote:
>On Thu, Jul 27, 2017 at 11:03:50AM +0200, Ludovic Courtès wrote:
>> Hi Efraim,
>> 
>> Efraim Flashner  skribis:
>> 
>> > There's a lot of perl related build failures. Maybe it would be
>better
>> > to revert the perl update and work on updating perl and all the
>perl
>> > modules separately. It seems to me that there are a large number of
>perl
>> > packages that haven't been updated in quite some time.
>> 
>> Reverting is not an option at this point IMO.  There are several
>Date::*
>> modules required by Biber that FTBFS and need an update, indeed, but
>I
>> think we should rather find a way to fix them (I spent a bit of time
>on
>> it but then moved on to something else.)
>> 
>> Thoughts?
>
>I'll work on building Biber now.
>
>Are there any other failing Perl modules that we *need* to fix? I think
>we can't achieve zero regressions from an update like this, especially
>since so many of these modules seem to lack an active upstream. So, if
>Guix users care about them, they should speak up now :)

Not as a reason to put it off, but Debian is beginning their Perl transition 
and I've noticed some of the modules I've looked at have Debian developers as 
the upstream.

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: revert perl-5.26.0 update?

2017-07-27 Thread Leo Famulari 
On Thu, Jul 27, 2017 at 11:03:50AM +0200, Ludovic Courtès wrote:
> Hi Efraim,
> 
> Efraim Flashner  skribis:
> 
> > There's a lot of perl related build failures. Maybe it would be better
> > to revert the perl update and work on updating perl and all the perl
> > modules separately. It seems to me that there are a large number of perl
> > packages that haven't been updated in quite some time.
> 
> Reverting is not an option at this point IMO.  There are several Date::*
> modules required by Biber that FTBFS and need an update, indeed, but I
> think we should rather find a way to fix them (I spent a bit of time on
> it but then moved on to something else.)
> 
> Thoughts?

I'll work on building Biber now.

Are there any other failing Perl modules that we *need* to fix? I think
we can't achieve zero regressions from an update like this, especially
since so many of these modules seem to lack an active upstream. So, if
Guix users care about them, they should speak up now :)


signature.asc
Description: PGP signature

Re: revert perl-5.26.0 update?

2017-07-27 Thread Leo Famulari 
On Thu, Jul 27, 2017 at 11:03:50AM +0200, Ludovic Courtès wrote:
> Hi Efraim,
> 
> Efraim Flashner  skribis:
> 
> > There's a lot of perl related build failures. Maybe it would be better
> > to revert the perl update and work on updating perl and all the perl
> > modules separately. It seems to me that there are a large number of perl
> > packages that haven't been updated in quite some time.
> 
> Reverting is not an option at this point IMO.  There are several Date::*
> modules required by Biber that FTBFS and need an update, indeed, but I
> think we should rather find a way to fix them (I spent a bit of time on
> it but then moved on to something else.)
> 
> Thoughts?

The "stereotypical" Perl 5.26.0-related failures are the "dotless @INC"
and regex brace-escaping issues. Both of these can be fixed downstream
(in Guix) fairly easily, if necessary.


signature.asc
Description: PGP signature

Re: revert perl-5.26.0 update?

2017-07-27 Thread Ludovic Courtès 
Hi Efraim,

Efraim Flashner  skribis:

> There's a lot of perl related build failures. Maybe it would be better
> to revert the perl update and work on updating perl and all the perl
> modules separately. It seems to me that there are a large number of perl
> packages that haven't been updated in quite some time.

Reverting is not an option at this point IMO.  There are several Date::*
modules required by Biber that FTBFS and need an update, indeed, but I
think we should rather find a way to fix them (I spent a bit of time on
it but then moved on to something else.)

Thoughts?

Ludo’.