Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Mon, 24 Oct 2016 17:15:24 +0200
Dominik 'Rathann' Mierzejewski  wrote:

> > > I was hoping to backport 2.0.0 to F25 as well, but I didn't have
> > > time to do it before Alpha or even Beta Freeze. Also, considering
> > > upstream has not released 2.0.0 yet, I guess I'll stick with
> > > 1.5.x on F25.
> > 
> > I've tried with python-matplotlib-1.5.2-0.3.rc2.fc25 [1] but it now
> > fails in the test-suite on x86_64 -
> > http://koji.fedoraproject.org/koji/taskinfo?taskID=16187221
> > Maybe we need to disable the test-suite globally ...
> 
> I spent quite a lot of effort on making the 2.0.0b4 tests pass on
> rawhide, so I think I'll do it for 1.5.x as well. Please open a bug
> and I'll get to it sometime soon.

yep, I saw the patches touching the test-suite

reported as https://bugzilla.redhat.com/show_bug.cgi?id=1388334


Dan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Monday, 24 October 2016 at 16:52, Dan Horák wrote:
> On Mon, 24 Oct 2016 14:49:29 +0200
> Dominik 'Rathann' Mierzejewski  wrote:
> 
> > On Monday, 24 October 2016 at 14:45, Dan Horák wrote:
> > > On Mon, 24 Oct 2016 14:30:42 +0200
> > > Dominik 'Rathann' Mierzejewski  wrote:
> > > 
> > > > On Monday, 24 October 2016 at 11:46, Dan Horák wrote:
> > > > > On Mon, 3 Oct 2016 17:57:47 +0200
> > > > > Dan Horák  wrote:
> > > > > 
> > > > > > On Mon, 3 Oct 2016 17:25:39 +0200
> > > > > > Dominik 'Rathann' Mierzejewski  wrote:
> > > > > > 
> > > > > > > On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> > > > [...]
> > > > > > > > Could we disable the test-suite for alternative arches
> > > > > > > > until there is proper upstream fix? Quite a number of
> > > > > > > > packages/builds are blocked on python-matplotlib
> > > > > > > > currently.
> > > > > > > 
> > > > > > > Which arches are we talking about? s390x, ppc64(le)?
> > > > > > > Anything else?
> > > > > > 
> > > > > > plus aarch64, question can be whether to make the list
> > > > > > "positive" %ifnarch arm, x86 then skip test results
> > > > > > or "negative" with %ifarch aarch64  ppc64 s390 ...
> > > > > 
> > > > > how about using
> > > > > 
> > > > > diff --git a/python-matplotlib.spec b/python-matplotlib.spec
> > > > > index 14d096b..bf30b5b 100644
> > > > > --- a/python-matplotlib.spec
> > > > > +++ b/python-matplotlib.spec
> > > > > @@ -5,7 +5,11 @@
> > > > >  %endif
> > > > >  %global __provides_exclude_from .*/site-packages/.*\\.so$
> > > > >  %global with_html   0
> > > > > +%ifarch %{power64} s390x
> > > > > +%global run_tests   0
> > > > > +%else
> > > > >  %global run_tests   1
> > > > > +%endif
> > > > >  
> > > > >  # On RHEL 7 onwards, don't build with wx:
> > > > >  %if 0%{?rhel} >= 7
> > > > > 
> > > > > until there is a better solution? Currently the missing
> > > > > python-matplotlib blocks a number of other packages and we need
> > > > > something before ppc koji will merge with primary.
> > > > 
> > > > That's fine, feel free to make this change yourself.
> > > 
> > > ok, new build will land in koji in a moment
> > 
> > Thank you.
> 
> done with python-matplotlib-2.0.0-0.3.b4.fc26
>  
> > > any objection if I would apply it to the f25 branch too?
> > 
> > No objection.
> > 
> > > I suppose it should stay on matplotlib 1.5.x
> > 
> > I was hoping to backport 2.0.0 to F25 as well, but I didn't have time
> > to do it before Alpha or even Beta Freeze. Also, considering upstream
> > has not released 2.0.0 yet, I guess I'll stick with 1.5.x on F25.
> 
> I've tried with python-matplotlib-1.5.2-0.3.rc2.fc25 [1] but it now
> fails in the test-suite on x86_64 -
> http://koji.fedoraproject.org/koji/taskinfo?taskID=16187221
> Maybe we need to disable the test-suite globally ...

I spent quite a lot of effort on making the 2.0.0b4 tests pass on rawhide,
so I think I'll do it for 1.5.x as well. Please open a bug and I'll get to
it sometime soon.

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Mon, 24 Oct 2016 14:49:29 +0200
Dominik 'Rathann' Mierzejewski  wrote:

> On Monday, 24 October 2016 at 14:45, Dan Horák wrote:
> > On Mon, 24 Oct 2016 14:30:42 +0200
> > Dominik 'Rathann' Mierzejewski  wrote:
> > 
> > > On Monday, 24 October 2016 at 11:46, Dan Horák wrote:
> > > > On Mon, 3 Oct 2016 17:57:47 +0200
> > > > Dan Horák  wrote:
> > > > 
> > > > > On Mon, 3 Oct 2016 17:25:39 +0200
> > > > > Dominik 'Rathann' Mierzejewski  wrote:
> > > > > 
> > > > > > On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> > > [...]
> > > > > > > Could we disable the test-suite for alternative arches
> > > > > > > until there is proper upstream fix? Quite a number of
> > > > > > > packages/builds are blocked on python-matplotlib
> > > > > > > currently.
> > > > > > 
> > > > > > Which arches are we talking about? s390x, ppc64(le)?
> > > > > > Anything else?
> > > > > 
> > > > > plus aarch64, question can be whether to make the list
> > > > > "positive" %ifnarch arm, x86 then skip test results
> > > > > or "negative" with %ifarch aarch64  ppc64 s390 ...
> > > > 
> > > > how about using
> > > > 
> > > > diff --git a/python-matplotlib.spec b/python-matplotlib.spec
> > > > index 14d096b..bf30b5b 100644
> > > > --- a/python-matplotlib.spec
> > > > +++ b/python-matplotlib.spec
> > > > @@ -5,7 +5,11 @@
> > > >  %endif
> > > >  %global __provides_exclude_from .*/site-packages/.*\\.so$
> > > >  %global with_html   0
> > > > +%ifarch %{power64} s390x
> > > > +%global run_tests   0
> > > > +%else
> > > >  %global run_tests   1
> > > > +%endif
> > > >  
> > > >  # On RHEL 7 onwards, don't build with wx:
> > > >  %if 0%{?rhel} >= 7
> > > > 
> > > > until there is a better solution? Currently the missing
> > > > python-matplotlib blocks a number of other packages and we need
> > > > something before ppc koji will merge with primary.
> > > 
> > > That's fine, feel free to make this change yourself.
> > 
> > ok, new build will land in koji in a moment
> 
> Thank you.

done with python-matplotlib-2.0.0-0.3.b4.fc26
 
> > any objection if I would apply it to the f25 branch too?
> 
> No objection.
> 
> > I suppose it should stay on matplotlib 1.5.x
> 
> I was hoping to backport 2.0.0 to F25 as well, but I didn't have time
> to do it before Alpha or even Beta Freeze. Also, considering upstream
> has not released 2.0.0 yet, I guess I'll stick with 1.5.x on F25.

I've tried with python-matplotlib-1.5.2-0.3.rc2.fc25 [1] but it now
fails in the test-suite on x86_64 -
http://koji.fedoraproject.org/koji/taskinfo?taskID=16187221
Maybe we need to disable the test-suite globally ...


Dan

[1]
http://pkgs.fedoraproject.org/cgit/rpms/python-matplotlib.git/commit/?h=f25=6afee769845272a53d23a5b01cb73bc1d06bfa80
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Monday, 24 October 2016 at 14:45, Dan Horák wrote:
> On Mon, 24 Oct 2016 14:30:42 +0200
> Dominik 'Rathann' Mierzejewski  wrote:
> 
> > On Monday, 24 October 2016 at 11:46, Dan Horák wrote:
> > > On Mon, 3 Oct 2016 17:57:47 +0200
> > > Dan Horák  wrote:
> > > 
> > > > On Mon, 3 Oct 2016 17:25:39 +0200
> > > > Dominik 'Rathann' Mierzejewski  wrote:
> > > > 
> > > > > On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> > [...]
> > > > > > Could we disable the test-suite for alternative arches until
> > > > > > there is proper upstream fix? Quite a number of
> > > > > > packages/builds are blocked on python-matplotlib currently.
> > > > > 
> > > > > Which arches are we talking about? s390x, ppc64(le)? Anything
> > > > > else?
> > > > 
> > > > plus aarch64, question can be whether to make the list "positive"
> > > > %ifnarch arm, x86 then skip test results
> > > > or "negative" with %ifarch aarch64  ppc64 s390 ...
> > > 
> > > how about using
> > > 
> > > diff --git a/python-matplotlib.spec b/python-matplotlib.spec
> > > index 14d096b..bf30b5b 100644
> > > --- a/python-matplotlib.spec
> > > +++ b/python-matplotlib.spec
> > > @@ -5,7 +5,11 @@
> > >  %endif
> > >  %global __provides_exclude_from .*/site-packages/.*\\.so$
> > >  %global with_html   0
> > > +%ifarch %{power64} s390x
> > > +%global run_tests   0
> > > +%else
> > >  %global run_tests   1
> > > +%endif
> > >  
> > >  # On RHEL 7 onwards, don't build with wx:
> > >  %if 0%{?rhel} >= 7
> > > 
> > > until there is a better solution? Currently the missing
> > > python-matplotlib blocks a number of other packages and we need
> > > something before ppc koji will merge with primary.
> > 
> > That's fine, feel free to make this change yourself.
> 
> ok, new build will land in koji in a moment

Thank you.

> any objection if I would apply it to the f25 branch too?

No objection.

> I suppose it should stay on matplotlib 1.5.x

I was hoping to backport 2.0.0 to F25 as well, but I didn't have time
to do it before Alpha or even Beta Freeze. Also, considering upstream
has not released 2.0.0 yet, I guess I'll stick with 1.5.x on F25.

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Mon, 24 Oct 2016 14:30:42 +0200
Dominik 'Rathann' Mierzejewski  wrote:

> On Monday, 24 October 2016 at 11:46, Dan Horák wrote:
> > On Mon, 3 Oct 2016 17:57:47 +0200
> > Dan Horák  wrote:
> > 
> > > On Mon, 3 Oct 2016 17:25:39 +0200
> > > Dominik 'Rathann' Mierzejewski  wrote:
> > > 
> > > > On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> [...]
> > > > > Could we disable the test-suite for alternative arches until
> > > > > there is proper upstream fix? Quite a number of
> > > > > packages/builds are blocked on python-matplotlib currently.
> > > > 
> > > > Which arches are we talking about? s390x, ppc64(le)? Anything
> > > > else?
> > > 
> > > plus aarch64, question can be whether to make the list "positive"
> > > %ifnarch arm, x86 then skip test results
> > > or "negative" with %ifarch aarch64  ppc64 s390 ...
> > 
> > how about using
> > 
> > diff --git a/python-matplotlib.spec b/python-matplotlib.spec
> > index 14d096b..bf30b5b 100644
> > --- a/python-matplotlib.spec
> > +++ b/python-matplotlib.spec
> > @@ -5,7 +5,11 @@
> >  %endif
> >  %global __provides_exclude_from .*/site-packages/.*\\.so$
> >  %global with_html   0
> > +%ifarch %{power64} s390x
> > +%global run_tests   0
> > +%else
> >  %global run_tests   1
> > +%endif
> >  
> >  # On RHEL 7 onwards, don't build with wx:
> >  %if 0%{?rhel} >= 7
> > 
> > until there is a better solution? Currently the missing
> > python-matplotlib blocks a number of other packages and we need
> > something before ppc koji will merge with primary.
> 
> That's fine, feel free to make this change yourself.

ok, new build will land in koji in a moment

any objection if I would apply it to the f25 branch too? I suppose it
should stay on matplotlib 1.5.x


Dan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Monday, 24 October 2016 at 11:46, Dan Horák wrote:
> On Mon, 3 Oct 2016 17:57:47 +0200
> Dan Horák  wrote:
> 
> > On Mon, 3 Oct 2016 17:25:39 +0200
> > Dominik 'Rathann' Mierzejewski  wrote:
> > 
> > > On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
[...]
> > > > Could we disable the test-suite for alternative arches until there
> > > > is proper upstream fix? Quite a number of packages/builds are
> > > > blocked on python-matplotlib currently.
> > > 
> > > Which arches are we talking about? s390x, ppc64(le)? Anything else?
> > 
> > plus aarch64, question can be whether to make the list "positive"
> > %ifnarch arm, x86 then skip test results
> > or "negative" with %ifarch aarch64  ppc64 s390 ...
> 
> how about using
> 
> diff --git a/python-matplotlib.spec b/python-matplotlib.spec
> index 14d096b..bf30b5b 100644
> --- a/python-matplotlib.spec
> +++ b/python-matplotlib.spec
> @@ -5,7 +5,11 @@
>  %endif
>  %global __provides_exclude_from .*/site-packages/.*\\.so$
>  %global with_html   0
> +%ifarch %{power64} s390x
> +%global run_tests   0
> +%else
>  %global run_tests   1
> +%endif
>  
>  # On RHEL 7 onwards, don't build with wx:
>  %if 0%{?rhel} >= 7
> 
> until there is a better solution? Currently the missing
> python-matplotlib blocks a number of other packages and we need
> something before ppc koji will merge with primary.

That's fine, feel free to make this change yourself.

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Mon, 3 Oct 2016 17:57:47 +0200
Dan Horák  wrote:

> On Mon, 3 Oct 2016 17:25:39 +0200
> Dominik 'Rathann' Mierzejewski  wrote:
> 
> > On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> > > On Thu, 22 Sep 2016 18:10:44 +0200
> > > Dominik 'Rathann' Mierzejewski  wrote:
> > > 
> > > > On Thursday, 22 September 2016 at 16:55, Dan Horák wrote:
> > > > > On Thu, 22 Sep 2016 15:41:01 +0200
> > > > > Dominik 'Rathann' Mierzejewski  wrote:
> > > > > 
> > > > > > Hello,
> > > > > > I've just pushed (but not built) python-matplotlib-2.0.0b4
> > > > > > to rawhide. I'll be attempting to rebuild all the affected
> > > > > > packages locally to test if they're compatible. In the
> > > > > > meantime, feel free to git pull and build locally for your
> > > > > > own testing.
> > > > > 
> > > > > python-matplotlib has been a pain for alternative arches
> > > > > recently, due failures in the test-suite (in 1.5.x). Is the
> > > > > 2.0 version going to be better in this regard?
> > > > 
> > > > tldr: no, but upstream is aware and working on it.
> > > > 
> > > > There are quite a few issues opened on upstream github tracker
> > > > either by me or by Debian maintainer. The main problem is that
> > > > upstream checks that rendering output matches their generated
> > > > images. This changes slightly with different versions of
> > > > freetype. As a workaround, I reintroduced a certain base
> > > > tolerance in order not to add specific tolerance value to each
> > > > failing test (there are ~900 failing due to slight rendering
> > > > differences). Then, there are some arch-specific issues, mainly
> > > > on ARM.
> > > 
> > > Could we disable the test-suite for alternative arches until there
> > > is proper upstream fix? Quite a number of packages/builds are
> > > blocked on python-matplotlib currently.
> > 
> > Which arches are we talking about? s390x, ppc64(le)? Anything else?
> 
> plus aarch64, question can be whether to make the list "positive"
> %ifnarch arm, x86 then skip test results
> or "negative" with %ifarch aarch64  ppc64 s390 ...

how about using

diff --git a/python-matplotlib.spec b/python-matplotlib.spec
index 14d096b..bf30b5b 100644
--- a/python-matplotlib.spec
+++ b/python-matplotlib.spec
@@ -5,7 +5,11 @@
 %endif
 %global __provides_exclude_from .*/site-packages/.*\\.so$
 %global with_html   0
+%ifarch %{power64} s390x
+%global run_tests   0
+%else
 %global run_tests   1
+%endif
 
 # On RHEL 7 onwards, don't build with wx:
 %if 0%{?rhel} >= 7

until there is a better solution? Currently the missing
python-matplotlib blocks a number of other packages and we need
something before ppc koji will merge with primary.


Dan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Christian Krause]

Hi,

On Thu, Sep 22, 2016 at 3:41 PM, Dominik 'Rathann' Mierzejewski <
domi...@greysector.net> wrote:

>
> I've just pushed (but not built) python-matplotlib-2.0.0b4 to rawhide.
> I'll be attempting to rebuild all the affected packages locally to
> test if they're compatible. In the meantime, feel free to git pull
> and build locally for your own testing.
>
> $ dnf repoquery --srpm --releasever=rawhide --whatrequires
> python-matplotlib --alldeps
>
> anki
>

Thanks for the heads-up. I just tested anki against
python-matplotlib-2.0.0b4: works fine, I haven't seen any issues.

There is no need to rebuild anki since it doesn't have a "BuildRequires:"
for python-matplotlib. The dependency is just caused by a "Requires:".

Best regards,
Christian
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Mon, 3 Oct 2016 17:25:39 +0200
Dominik 'Rathann' Mierzejewski  wrote:

> On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> > On Thu, 22 Sep 2016 18:10:44 +0200
> > Dominik 'Rathann' Mierzejewski  wrote:
> > 
> > > On Thursday, 22 September 2016 at 16:55, Dan Horák wrote:
> > > > On Thu, 22 Sep 2016 15:41:01 +0200
> > > > Dominik 'Rathann' Mierzejewski  wrote:
> > > > 
> > > > > Hello,
> > > > > I've just pushed (but not built) python-matplotlib-2.0.0b4 to
> > > > > rawhide. I'll be attempting to rebuild all the affected
> > > > > packages locally to test if they're compatible. In the
> > > > > meantime, feel free to git pull and build locally for your
> > > > > own testing.
> > > > 
> > > > python-matplotlib has been a pain for alternative arches
> > > > recently, due failures in the test-suite (in 1.5.x). Is the 2.0
> > > > version going to be better in this regard?
> > > 
> > > tldr: no, but upstream is aware and working on it.
> > > 
> > > There are quite a few issues opened on upstream github tracker
> > > either by me or by Debian maintainer. The main problem is that
> > > upstream checks that rendering output matches their generated
> > > images. This changes slightly with different versions of
> > > freetype. As a workaround, I reintroduced a certain base
> > > tolerance in order not to add specific tolerance value to each
> > > failing test (there are ~900 failing due to slight rendering
> > > differences). Then, there are some arch-specific issues, mainly
> > > on ARM.
> > 
> > Could we disable the test-suite for alternative arches until there
> > is proper upstream fix? Quite a number of packages/builds are
> > blocked on python-matplotlib currently.
> 
> Which arches are we talking about? s390x, ppc64(le)? Anything else?

plus aarch64, question can be whether to make the list "positive"
%ifnarch arm, x86 then skip test results
or "negative" with %ifarch aarch64  ppc64 s390 ...

> 
> > For math precision issues -ffp-contract=off often helps. And yes,
> > comparing rendered output can be considered fragile, we have already
> > met it ...
> 
> Where does that parameter go?

into CFLAGS, it's for gcc


Dan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Monday, 03 October 2016 at 10:43, Dan Horák wrote:
> On Thu, 22 Sep 2016 18:10:44 +0200
> Dominik 'Rathann' Mierzejewski  wrote:
> 
> > On Thursday, 22 September 2016 at 16:55, Dan Horák wrote:
> > > On Thu, 22 Sep 2016 15:41:01 +0200
> > > Dominik 'Rathann' Mierzejewski  wrote:
> > > 
> > > > Hello,
> > > > I've just pushed (but not built) python-matplotlib-2.0.0b4 to
> > > > rawhide. I'll be attempting to rebuild all the affected packages
> > > > locally to test if they're compatible. In the meantime, feel free
> > > > to git pull and build locally for your own testing.
> > > 
> > > python-matplotlib has been a pain for alternative arches recently,
> > > due failures in the test-suite (in 1.5.x). Is the 2.0 version going
> > > to be better in this regard?
> > 
> > tldr: no, but upstream is aware and working on it.
> > 
> > There are quite a few issues opened on upstream github tracker either
> > by me or by Debian maintainer. The main problem is that upstream
> > checks that rendering output matches their generated images. This
> > changes slightly with different versions of freetype. As a
> > workaround, I reintroduced a certain base tolerance in order not to
> > add specific tolerance value to each failing test (there are ~900
> > failing due to slight rendering differences). Then, there are some
> > arch-specific issues, mainly on ARM.
> 
> Could we disable the test-suite for alternative arches until there is
> proper upstream fix? Quite a number of packages/builds are blocked on
> python-matplotlib currently.

Which arches are we talking about? s390x, ppc64(le)? Anything else?

> For math precision issues -ffp-contract=off often helps. And yes,
> comparing rendered output can be considered fragile, we have already
> met it ...

Where does that parameter go?

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Thu, 22 Sep 2016 18:10:44 +0200
Dominik 'Rathann' Mierzejewski  wrote:

> On Thursday, 22 September 2016 at 16:55, Dan Horák wrote:
> > On Thu, 22 Sep 2016 15:41:01 +0200
> > Dominik 'Rathann' Mierzejewski  wrote:
> > 
> > > Hello,
> > > I've just pushed (but not built) python-matplotlib-2.0.0b4 to
> > > rawhide. I'll be attempting to rebuild all the affected packages
> > > locally to test if they're compatible. In the meantime, feel free
> > > to git pull and build locally for your own testing.
> > 
> > python-matplotlib has been a pain for alternative arches recently,
> > due failures in the test-suite (in 1.5.x). Is the 2.0 version going
> > to be better in this regard?
> 
> tldr: no, but upstream is aware and working on it.
> 
> There are quite a few issues opened on upstream github tracker either
> by me or by Debian maintainer. The main problem is that upstream
> checks that rendering output matches their generated images. This
> changes slightly with different versions of freetype. As a
> workaround, I reintroduced a certain base tolerance in order not to
> add specific tolerance value to each failing test (there are ~900
> failing due to slight rendering differences). Then, there are some
> arch-specific issues, mainly on ARM.

Could we disable the test-suite for alternative arches until there is
proper upstream fix? Quite a number of packages/builds are blocked on
python-matplotlib currently.

For math precision issues -ffp-contract=off often helps. And yes,
comparing rendered output can be considered fragile, we have already
met it ...


Dan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Monday, 26 September 2016 at 13:58, Dominik 'Rathann' Mierzejewski wrote:
> On Thursday, 22 September 2016 at 15:41, Dominik 'Rathann' Mierzejewski wrote:
> > Hello,
> > I've just pushed (but not built) python-matplotlib-2.0.0b4 to rawhide.
> > I'll be attempting to rebuild all the affected packages locally to
> > test if they're compatible. In the meantime, feel free to git pull
> > and build locally for your own testing.
> > 
> > $ dnf repoquery --srpm --releasever=rawhide --whatrequires 
> > python-matplotlib --alldeps
> 
> Most packages rebuilt locally without errors, so
> python-matplotlib-2.0.0-0.1.b4.fc26 is currently building in koji.
> 
> Here are the problematic ones:
> 
> > APLpy
> 
> Rebuild failed with (python3):
> AttributeError: module 'distutils.config' has no attribute 'ConfigParser'

This was actually an upstream bug, reported and fixed already:
https://github.com/aplpy/aplpy/issues/304
Then, there were dependency issues - (Build)Requires: python-astropy
brings the python3 version instead of the expected python2 version.
Bug filed: https://bugzilla.redhat.com/show_bug.cgi?id=1380135 .
And, finally, there are some test failures, but I'm unable to tell
if this is related to matplotlib2 or not. Bug filed:
https://bugzilla.redhat.com/show_bug.cgi?id=1380134 .

> > python-igor
> 
> Looks like it got built with python3 and fails %check grepping for
> python2.

Fixed, but it should switch to py3 by default.

> > python-nipy
> 
> Missing (Build)dependency on python2-functools32. Also, 3 genuine test
> failures:
> ValueError: object arrays are not supported

Known issue, already reported upstream by maintainer. Bug filed:
https://bugzilla.redhat.com/show_bug.cgi?id=1380126

> > python-pebl
> 
> mv: 'docs/src/install.rst' and 'INSTALL.txt' are the same file

Fixed.

> > python-pyriemann
> 
> Missing (Build)dependency on python2-functools32.
> 
> > python-seaborn
> 
> Missing (Build)dependency on python2-functools32.

Fixed by adding the dependency to matplotlib package where it belongs.

Additionally, I had to add ExcludeArch: aarch64 to sagemath due to
missing dependencies on aarch64:
https://bugzilla.redhat.com/show_bug.cgi?id=1380102
https://bugzilla.redhat.com/show_bug.cgi?id=1380101

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Thursday, 22 September 2016 at 15:41, Dominik 'Rathann' Mierzejewski wrote:
> Hello,
> I've just pushed (but not built) python-matplotlib-2.0.0b4 to rawhide.
> I'll be attempting to rebuild all the affected packages locally to
> test if they're compatible. In the meantime, feel free to git pull
> and build locally for your own testing.
> 
> $ dnf repoquery --srpm --releasever=rawhide --whatrequires python-matplotlib 
> --alldeps

Most packages rebuilt locally without errors, so
python-matplotlib-2.0.0-0.1.b4.fc26 is currently building in koji.

Here are the problematic ones:

> APLpy

Rebuild failed with (python3):
AttributeError: module 'distutils.config' has no attribute 'ConfigParser'

> python-igor

Looks like it got built with python3 and fails %check grepping for
python2.

> python-nipy

Missing (Build)dependency on python2-functools32. Also, 3 genuine test
failures:
ValueError: object arrays are not supported

> python-pebl

mv: 'docs/src/install.rst' and 'INSTALL.txt' are the same file

> python-pyriemann

Missing (Build)dependency on python2-functools32.

> python-seaborn

Missing (Build)dependency on python2-functools32.

I'll attempt to fix the above and rebuild over the coming days.

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dominik 'Rathann' Mierzejewski]

On Thursday, 22 September 2016 at 16:55, Dan Horák wrote:
> On Thu, 22 Sep 2016 15:41:01 +0200
> Dominik 'Rathann' Mierzejewski  wrote:
> 
> > Hello,
> > I've just pushed (but not built) python-matplotlib-2.0.0b4 to rawhide.
> > I'll be attempting to rebuild all the affected packages locally to
> > test if they're compatible. In the meantime, feel free to git pull
> > and build locally for your own testing.
> 
> python-matplotlib has been a pain for alternative arches recently, due
> failures in the test-suite (in 1.5.x). Is the 2.0 version going to be
> better in this regard?

tldr: no, but upstream is aware and working on it.

There are quite a few issues opened on upstream github tracker either
by me or by Debian maintainer. The main problem is that upstream checks
that rendering output matches their generated images. This changes
slightly with different versions of freetype. As a workaround, I
reintroduced a certain base tolerance in order not to add specific
tolerance value to each failing test (there are ~900 failing due to
slight rendering differences). Then, there are some arch-specific
issues, mainly on ARM.

The update is necessary as there are no more releases planned for
the 1.5.x branch.

Regards,
Dominik
-- 
Fedora http://fedoraproject.org/wiki/User:Rathann
RPMFusion http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org

Re: [HEADS UP] python-matplotlib-2.0.0 major update

[Dan Horák]

On Thu, 22 Sep 2016 15:41:01 +0200
Dominik 'Rathann' Mierzejewski  wrote:

> Hello,
> I've just pushed (but not built) python-matplotlib-2.0.0b4 to rawhide.
> I'll be attempting to rebuild all the affected packages locally to
> test if they're compatible. In the meantime, feel free to git pull
> and build locally for your own testing.

python-matplotlib has been a pain for alternative arches recently, due
failures in the test-suite (in 1.5.x). Is the 2.0 version going to be
better in this regard?


Dan
___
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org