matplotlib-devel Mailing List for matplotlib

Hello,

This may be a silly question, but I'm wondering what happens in
Matplotlib rendering when there is a "big" Line2D object (say 10**7
points) added to an Axes, with *visibility set to false*.

Here is an tiny script meant to be run interactively (say with Python in
pylab mode) step by step :
https://gist.github.com/3638471

I'm experiencing a strange slowdown of Axes rendering when such a "big
line" is added to the Axes, despite the fact it is invisible. So I'm
wondering if this is an expected behavior or a some kind of bug. Or
maybe it's a well known fact, already fixed in the upcoming 1.2 release
and in that case sorry for the duplicate.

Best,
Pierre

PS: I'm using Matplotlib version 1.1.1rc2, from Debian Testing. Today
TkAgg backend, but the other day same problem happened with Qt.

On 9/10/2012 10:08 AM, Michael Droettboom wrote:
> On 09/10/2012 12:49 PM, Mark Lawrence wrote:
>> My offer to test on Windows still holds.  The question is test what?
>> Assuming that the intent is to support Python 2.6/7 and 3.1/2/3 then
>> different versions of Visual Studio are needed as detailed here
>> http://bugs.python.org/issue13210.  I can't see much sense in me trying
>> to build and test all that lot while other volunteers are doing exactly
>> the same thing at the same time.  Is there a cunning plan somewhere to
>> cover this that I've missed?
>>
>
> The process is usually that the release is tagged in github, a tarball
> is created, and then various platform specialists create the binary
> releases.  For the last release Christoph Gohlke did the Windows builds,
> and I assume he is able to do that again this time. Perhaps it makes
> sense to coordinate with him to share some of that work?  Anything we
> can do to lessen the burden on any one person is always appreciated.
>
> Even leading up to the tagging of the release candidate, it's always
> helpful to build from github periodically and test on the platform and
> version of Python that's most important to you, and report back into the
> issue tracker any test failures or other glitches.
>
> Mike
>


Hello,

sure, I can provide the Windows binaries and run tests. I have not tried 
recent master, but I am using a ~2 month old build on Python 3.2 and 3.3 
without any issues. As for Python 3.3, there were no problems building 
matplotlib with Visual Studio 2010, but there is no official numpy 
release supporting Python 3.3 yet. I also have a PIL fork that works on 
Python 3, which is good for testing.

Christoph

On Mon, Sep 10, 2012 at 5:00 PM, Michael Droettboom <md...@st...> wrote:
> We have a few outstanding issues, which I'll categorize below:
>
> Critical things that need just a little more work:
>
> #1223    dpi= for bitmaps not handled correctly

I don't believe this is release critical. It only affects metadata.
Eric has fixed this for TIFFs but there may be outstanding issues with
other bitmap types. My feeling is that the below issues should receive
more attention.

> #1209    Pass linewidth to Mac context properly
> #786    savefig() renders paths and text differently than show()
> #113    dpi= doesn't seem to have any effect with MacOS X backend

Michiel has provided great feedback regarding the previously proposed
solution to part of this problem, and this has been reflected in the
updated pull request.

-- 
Damon McDougall
http://www.damon.is-a-geek.com
B2.39
Mathematics Institute
University of Warwick
Coventry
West Midlands
CV4 7AL
United Kingdom

On 09/10/2012 12:49 PM, Mark Lawrence wrote:
> My offer to test on Windows still holds.  The question is test what?
> Assuming that the intent is to support Python 2.6/7 and 3.1/2/3 then
> different versions of Visual Studio are needed as detailed here
> http://bugs.python.org/issue13210.  I can't see much sense in me trying
> to build and test all that lot while other volunteers are doing exactly
> the same thing at the same time.  Is there a cunning plan somewhere to
> cover this that I've missed?
>

The process is usually that the release is tagged in github, a tarball 
is created, and then various platform specialists create the binary 
releases.  For the last release Christoph Gohlke did the Windows builds, 
and I assume he is able to do that again this time. Perhaps it makes 
sense to coordinate with him to share some of that work?  Anything we 
can do to lessen the burden on any one person is always appreciated.

Even leading up to the tagging of the release candidate, it's always 
helpful to build from github periodically and test on the platform and 
version of Python that's most important to you, and report back into the 
issue tracker any test failures or other glitches.

Mike

On 10/09/2012 17:00, Michael Droettboom wrote:
> Today is the scheduled day for release candidate 1.2rc1.
>
> We seem to be in really good shape.  Thanks to everyone that has been
> working so hard to squash bugs, particularly ones that turned out to be
> bottomless rabbit holes.
>
> We have a few outstanding issues, which I'll categorize below:
>
> Critical things that need just a little more work:
>
> #1223    dpi= for bitmaps not handled correctly
> #1209    Pass linewidth to Mac context properly
> #786    savefig() renders paths and text differently than show()
> #113    dpi= doesn't seem to have any effect with MacOS X backend
>
> #1208    FAIL: matplotlib.tests.test_text.test_contains.test
> release_critical
> #1176    Reverted a previous change to artist transform setting. Fixes
> legend bug.
>
> Non-critical features that are near completion -- just needing some
> confirmation or testing:
>
> #847    Add stacked kwarg to hist and implement stacked hists for step
> histtype
> #751    Building on osx with python 3.2 OSX
>
> Problems requiring a fix in another project:
>
> #1126    Qt4 save dialog not functional on CentOS-5
>
> Things that are probably too big to fix now, but deserve warnings in the
> release notes:
>
> #854    Bug in Axes.relim when the first line is y_isdata=False and
> possible fix
> #740    plt.pcolormesh and shape mismatch
> #162    twinx and plot_date confirmed
>
> Reminders for the release manager:
>
> #1207    Add contributor and git stats to documentation
> #1070    Use github for downloads
>
> I will write prose for the "too big to fix now" stuff, and I trust the
> other close things will get done by their respective authors. Then we're
> very close to getting a release candidate out.
>
> Congratulations to everyone for a job well done!
>
> Mike
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>

My offer to test on Windows still holds.  The question is test what? 
Assuming that the intent is to support Python 2.6/7 and 3.1/2/3 then 
different versions of Visual Studio are needed as detailed here 
http://bugs.python.org/issue13210.  I can't see much sense in me trying 
to build and test all that lot while other volunteers are doing exactly 
the same thing at the same time.  Is there a cunning plan somewhere to 
cover this that I've missed?

-- 
Cheers.

Mark Lawrence.

I'd certainly like to see that work continue, but I don't know if it's 
worth holding up the release candidate for.  We probably won't get it 
out today given the other critical things yet to go in -- but once the 
release candidate is cut, I'd prefer to be really conservative about 
what changes go in before the final release.  We can continue your great 
PEP8 work on master (after the 1.2.x maintenance branch is created), of 
course.

Mike

On 09/10/2012 12:10 PM, Nelle Varoquaux wrote:
> Hello,
>
> Do you still accept pep8 cleaning up ? I've got a couple of important 
> deadlines coming up soon, but I might be able to clean up the whole code.
>
> Thanks,
> N
>
> On 10 September 2012 18:00, Michael Droettboom <md...@st... 
> <mailto:md...@st...>> wrote:
>
>     Today is the scheduled day for release candidate 1.2rc1.
>
>     We seem to be in really good shape.  Thanks to everyone that has been
>     working so hard to squash bugs, particularly ones that turned out
>     to be
>     bottomless rabbit holes.
>
>     We have a few outstanding issues, which I'll categorize below:
>
>     Critical things that need just a little more work:
>
>     #1223    dpi= for bitmaps not handled correctly
>     #1209    Pass linewidth to Mac context properly
>     #786    savefig() renders paths and text differently than show()
>     #113    dpi= doesn't seem to have any effect with MacOS X backend
>
>     #1208    FAIL: matplotlib.tests.test_text.test_contains.test
>     release_critical
>     #1176    Reverted a previous change to artist transform setting. Fixes
>     legend bug.
>
>     Non-critical features that are near completion -- just needing some
>     confirmation or testing:
>
>     #847    Add stacked kwarg to hist and implement stacked hists for step
>     histtype
>     #751    Building on osx with python 3.2 OSX
>
>     Problems requiring a fix in another project:
>
>     #1126    Qt4 save dialog not functional on CentOS-5
>
>     Things that are probably too big to fix now, but deserve warnings
>     in the
>     release notes:
>
>     #854    Bug in Axes.relim when the first line is y_isdata=False and
>     possible fix
>     #740    plt.pcolormesh and shape mismatch
>     #162    twinx and plot_date confirmed
>
>     Reminders for the release manager:
>
>     #1207    Add contributor and git stats to documentation
>     #1070    Use github for downloads
>
>     I will write prose for the "too big to fix now" stuff, and I trust the
>     other close things will get done by their respective authors. Then
>     we're
>     very close to getting a release candidate out.
>
>     Congratulations to everyone for a job well done!
>
>     Mike
>
>     ------------------------------------------------------------------------------
>     Live Security Virtual Conference
>     Exclusive live event will cover all the ways today's security and
>     threat landscape has changed and how IT managers can respond.
>     Discussions
>     will include endpoint security, mobile security and the latest in
>     malware
>     threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>     _______________________________________________
>     Matplotlib-devel mailing list
>     Mat...@li...
>     <mailto:Mat...@li...>
>     https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>
>

Hello,

Do you still accept pep8 cleaning up ? I've got a couple of important
deadlines coming up soon, but I might be able to clean up the whole code.

Thanks,
N

On 10 September 2012 18:00, Michael Droettboom <md...@st...> wrote:

> Today is the scheduled day for release candidate 1.2rc1.
>
> We seem to be in really good shape.  Thanks to everyone that has been
> working so hard to squash bugs, particularly ones that turned out to be
> bottomless rabbit holes.
>
> We have a few outstanding issues, which I'll categorize below:
>
> Critical things that need just a little more work:
>
> #1223    dpi= for bitmaps not handled correctly
> #1209    Pass linewidth to Mac context properly
> #786    savefig() renders paths and text differently than show()
> #113    dpi= doesn't seem to have any effect with MacOS X backend
>
> #1208    FAIL: matplotlib.tests.test_text.test_contains.test
> release_critical
> #1176    Reverted a previous change to artist transform setting. Fixes
> legend bug.
>
> Non-critical features that are near completion -- just needing some
> confirmation or testing:
>
> #847    Add stacked kwarg to hist and implement stacked hists for step
> histtype
> #751    Building on osx with python 3.2 OSX
>
> Problems requiring a fix in another project:
>
> #1126    Qt4 save dialog not functional on CentOS-5
>
> Things that are probably too big to fix now, but deserve warnings in the
> release notes:
>
> #854    Bug in Axes.relim when the first line is y_isdata=False and
> possible fix
> #740    plt.pcolormesh and shape mismatch
> #162    twinx and plot_date confirmed
>
> Reminders for the release manager:
>
> #1207    Add contributor and git stats to documentation
> #1070    Use github for downloads
>
> I will write prose for the "too big to fix now" stuff, and I trust the
> other close things will get done by their respective authors. Then we're
> very close to getting a release candidate out.
>
> Congratulations to everyone for a job well done!
>
> Mike
>
>
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and
> threat landscape has changed and how IT managers can respond. Discussions
> will include endpoint security, mobile security and the latest in malware
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>

Today is the scheduled day for release candidate 1.2rc1.

We seem to be in really good shape.  Thanks to everyone that has been 
working so hard to squash bugs, particularly ones that turned out to be 
bottomless rabbit holes.

We have a few outstanding issues, which I'll categorize below:

Critical things that need just a little more work:

#1223    dpi= for bitmaps not handled correctly
#1209    Pass linewidth to Mac context properly
#786    savefig() renders paths and text differently than show()
#113    dpi= doesn't seem to have any effect with MacOS X backend

#1208    FAIL: matplotlib.tests.test_text.test_contains.test 
release_critical
#1176    Reverted a previous change to artist transform setting. Fixes 
legend bug.

Non-critical features that are near completion -- just needing some 
confirmation or testing:

#847    Add stacked kwarg to hist and implement stacked hists for step 
histtype
#751    Building on osx with python 3.2 OSX

Problems requiring a fix in another project:

#1126    Qt4 save dialog not functional on CentOS-5

Things that are probably too big to fix now, but deserve warnings in the 
release notes:

#854    Bug in Axes.relim when the first line is y_isdata=False and 
possible fix
#740    plt.pcolormesh and shape mismatch
#162    twinx and plot_date confirmed

Reminders for the release manager:

#1207    Add contributor and git stats to documentation
#1070    Use github for downloads

I will write prose for the "too big to fix now" stuff, and I trust the 
other close things will get done by their respective authors. Then we're 
very close to getting a release candidate out.

Congratulations to everyone for a job well done!

Mike