matplotlib-devel Mailing List for matplotlib (Page 2)

On Thu, Jun 14, 2012 at 11:08 AM, tcb <the...@gm...> wrote:
> Hi,
>
> I'm trying to plot a grid of images with no spaces in between them and I
> have problems with the snapping. The grid is computed in data coordinates
> and the positions of the images are determined by the extents parameter
>
> ax.imshow(read_png(filename), extent=extent, snap=True)
>
>
> The images and extents are always the same size and each image starts
> where the last finished
>
> (x0, x0+dx, y0, y0+dy)
> (x0+dx, x0+2*dx, y0, y0+dy)
> etc...
>
>
> but it seems something happens in the backend (both QT4 and GTKAgg) and
> pixel seams are visible between the images at various zoom levels.
>
> It seems the ImageGrid was developed to fix this kind of problem, but I'm
> not sure I can adapt my plot to use ImageGrid. I want to draw stuff over my
> images and I need to position them in data coordinates.
>
> I've tried the snap=True parameter to imshow and it doesn't seem to affect
> the output.
>
> Does anyone have a simple solution to this problem which eliminates the
> seams?
>
> thanks,
>
>
>
Could you try out the latest version of matplotlib from the master branch?
There have been some work in this area in the past few months. I don't
know if any of that work went into the recently announced v1.1.1-rc2.
Ben Root

Hi,
I'm trying to plot a grid of images with no spaces in between them and I
have problems with the snapping. The grid is computed in data coordinates
and the positions of the images are determined by the extents parameter
ax.imshow(read_png(filename), extent=extent, snap=True)
The images and extents are always the same size and each image starts where
the last finished
(x0, x0+dx, y0, y0+dy)
(x0+dx, x0+2*dx, y0, y0+dy)
etc...
but it seems something happens in the backend (both QT4 and GTKAgg) and
pixel seams are visible between the images at various zoom levels.
It seems the ImageGrid was developed to fix this kind of problem, but I'm
not sure I can adapt my plot to use ImageGrid. I want to draw stuff over my
images and I need to position them in data coordinates.
I've tried the snap=True parameter to imshow and it doesn't seem to affect
the output.
Does anyone have a simple solution to this problem which eliminates the
seams?
thanks,

I don't think this has anything to do with matplotlib, but instead just 
with pygobject (or perhaps just gtk itself) on Fedora 17.
I suspect that this fedora bug is responsible:
https://bugzilla.redhat.com/show_bug.cgi?id=790053
Importing gtk is enough to produce this warning:
In [1]: import gtk
** (process:15013): WARNING **: Trying to register gtype 
'GMountMountFlags' as enum when in fact it is of type 'GFlags'
** (process:15013): WARNING **: Trying to register gtype 
'GDriveStartFlags' as enum when in fact it is of type 'GFlags'
** (process:15013): WARNING **: Trying to register gtype 
'GSocketMsgFlags' as enum when in fact it is of type 'GFlags'
Mike
On 06/13/2012 05:40 PM, Neal Becker wrote:
> Since installing this on fedora f17 x86_64, I see this warning:
>
> ipython -pylab
> ...
>
> ** (process:26729): WARNING **: Trying to register gtype 'GMountMountFlags' as
> enum when in fact it is of type 'GFlags'
>
> ** (process:26729): WARNING **: Trying to register gtype 'GDriveStartFlags' as
> enum when in fact it is of type 'GFlags'
>
> ** (process:26729): WARNING **: Trying to register gtype 'GSocketMsgFlags' as
> enum when in fact it is of type 'GFlags'
>
> Welcome to pylab, a matplotlib-based Python environment [backend: GTKAgg].
> For more information, type 'help(pylab)'.
>
>
> ------------------------------------------------------------------------------
> 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

Since installing this on fedora f17 x86_64, I see this warning:
ipython -pylab
...
** (process:26729): WARNING **: Trying to register gtype 'GMountMountFlags' as 
enum when in fact it is of type 'GFlags'
** (process:26729): WARNING **: Trying to register gtype 'GDriveStartFlags' as 
enum when in fact it is of type 'GFlags'
** (process:26729): WARNING **: Trying to register gtype 'GSocketMsgFlags' as 
enum when in fact it is of type 'GFlags'
Welcome to pylab, a matplotlib-based Python environment [backend: GTKAgg].
For more information, type 'help(pylab)'.

On Wed, Jun 13, 2012 at 1:03 PM, Michael Droettboom <md...@st...> wrote:
> Is there a slight mix-up in the commits? In the email John sent out, it had
> commit f763cd11f50437568f8146b822ca8f25647cbb61 at the top of the log
> (alternative baseline image for mathfont_stix_14.png).
The tag is correct, the email log had an extra commit in it that is
not in the rc. My bad.

On 06/13/2012 01:29 PM, Benjamin Root wrote:
>
>
> On Mon, Jun 11, 2012 at 2:23 PM, Michael Droettboom <md...@st... 
> <mailto:md...@st...>> wrote:
>
> On 06/11/2012 02:17 PM, John Hunter wrote:
> > On Mon, Jun 11, 2012 at 10:30 AM, Michael
> Droettboom<md...@st... <mailto:md...@st...>> wrote:
> >> John: Do you know which git revision you made the tarballs
> from? I'll go
> >> ahead and tag it.
> >
> https://github.com/matplotlib/matplotlib/commit/97f713098a71c70d70c46602794c2a0fd126b614
> Thanks. Done.
>
> Here's what it was, if curious:
>
> >git tag v1.1.1-rc2 97f713098a71c70d70c46602794c2a0fd126b614
> >git push --tags upstream
> Total 0 (delta 0), reused 0 (delta 0)
> To gi...@gi...:matplotlib/matplotlib.git
> * [new tag] v1.1.1-rc2 -> v1.1.1-rc2
>
> Mike
>
>
> Is there a slight mix-up in the commits? In the email John sent out, 
> it had commit f763cd11f50437568f8146b822ca8f25647cbb61 at the top of 
> the log (alternative baseline image for mathfont_stix_14.png).
>
> Just checking.
Which e-mail is this? I couldn't figure out what commit the tarball was 
based on which is why I asked John. If we can confirm the correct 
commit, I'm happy to move the tag.
Mike

On Mon, Jun 11, 2012 at 2:23 PM, Michael Droettboom <md...@st...> wrote:
> On 06/11/2012 02:17 PM, John Hunter wrote:
> > On Mon, Jun 11, 2012 at 10:30 AM, Michael Droettboom<md...@st...>
> wrote:
> >> John: Do you know which git revision you made the tarballs from? I'll
> go
> >> ahead and tag it.
> >
> https://github.com/matplotlib/matplotlib/commit/97f713098a71c70d70c46602794c2a0fd126b614
> Thanks. Done.
>
> Here's what it was, if curious:
>
> >git tag v1.1.1-rc2 97f713098a71c70d70c46602794c2a0fd126b614
> >git push --tags upstream
> Total 0 (delta 0), reused 0 (delta 0)
> To gi...@gi...:matplotlib/matplotlib.git
> * [new tag] v1.1.1-rc2 -> v1.1.1-rc2
>
> Mike
>
>
Is there a slight mix-up in the commits? In the email John sent out, it
had commit f763cd11f50437568f8146b822ca8f25647cbb61 at the top of the log
(alternative baseline image for mathfont_stix_14.png).
Just checking.
Ben Root

On Sun, Jun 10, 2012 at 5:24 AM, Jouni K. Seppänen <jk...@ik...> wrote:
> Alan Griffiths <ala...@gm...> writes:
>
>> I've come across a problem (possibly a bug in cleanup_path) after
>> plotting a line containing nan's and then scaling the axes so that the
>> line is clipped by the axes.
>>
>> Instead of a gap at the position of the nan value, two extra segments
>> appear - back to the start and then onto the next-plus-one valid
>> point.
>
> This sounds a lot like
>
> https://github.com/matplotlib/matplotlib/issues/804
>
Thanks, yes it's fixed in 1.1.1rc2

On 06/11/2012 02:17 PM, John Hunter wrote:
> On Mon, Jun 11, 2012 at 10:30 AM, Michael Droettboom<md...@st...> wrote:
>> John: Do you know which git revision you made the tarballs from? I'll go
>> ahead and tag it.
> https://github.com/matplotlib/matplotlib/commit/97f713098a71c70d70c46602794c2a0fd126b614
Thanks. Done.
Here's what it was, if curious:
 >git tag v1.1.1-rc2 97f713098a71c70d70c46602794c2a0fd126b614
 >git push --tags upstream
Total 0 (delta 0), reused 0 (delta 0)
To gi...@gi...:matplotlib/matplotlib.git
 * [new tag] v1.1.1-rc2 -> v1.1.1-rc2
Mike

On Mon, Jun 11, 2012 at 10:30 AM, Michael Droettboom <md...@st...> wrote:
> John: Do you know which git revision you made the tarballs from? I'll go
> ahead and tag it.
https://github.com/matplotlib/matplotlib/commit/97f713098a71c70d70c46602794c2a0fd126b614

On Mon, Jun 11, 2012 at 3:40 PM, RuiDC <ru...@ya...> wrote:
>
> Testing looks good only two known fails for me against the rc2 binaries that
> I spotted on SF on both Windows7 amd64 and Windows XP x86_32 (both Python
> 2.7.3).
>
> The only thing out of the ordinary for me is the following on both:
>
> matplotlib.tests.test_axes.test_markevery_line.test ... ok
> C:\Python27\lib\site-packages\matplotlib\lines.py:461: RuntimeWarning:
> invalid value encountered in greater_equal
> return np.alltrue(x[1:]-x[0:-1]>=0)
> matplotlib.tests.test_axes.test_nonfinite_limits.test ... ok
>
> I'd seen this on Sandro Tosi's output on this ML, is this a known numpy
> issue? (using the Christoph Gohlke 1.6.2 MKL binaries)
yep: http://mail.scipy.org/pipermail/numpy-discussion/2012-June/062734.html
-- 
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

John: Do you know which git revision you made the tarballs from? I'll 
go ahead and tag it.
Mike
On 06/09/2012 09:52 PM, Benjamin Root wrote:
> Don't forget to tag the rc in git!
>
> On Sat, Jun 9, 2012 at 6:40 PM, John Hunter <jd...@gm... 
> <mailto:jd...@gm...>> wrote:
>
> On Sat, Jun 9, 2012 at 5:33 PM, John Hunter <jd...@gm...
> <mailto:jd...@gm...>> wrote:
> > On Sat, Jun 9, 2012 at 5:27 PM, Eric Firing <ef...@ha...
> <mailto:ef...@ha...>> wrote:
> >> It also has backend_gdk.c (supposed to be a temporary copy of
> >> _backend_gdk.c) and some others like it.
> >
> > OK, I'll rebuild them super clean... I know what went wrong.
>
> Just uploaded two new files built from clean git checkouts.
>
> ------------------------------------------------------------------------------
> 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
>
>
>
>
> ------------------------------------------------------------------------------
> 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

Testing looks good only two known fails for me against the rc2 binaries that
I spotted on SF on both Windows7 amd64 and Windows XP x86_32 (both Python
2.7.3).
The only thing out of the ordinary for me is the following on both:
matplotlib.tests.test_axes.test_markevery_line.test ... ok
C:\Python27\lib\site-packages\matplotlib\lines.py:461: RuntimeWarning:
invalid value encountered in greater_equal
 return np.alltrue(x[1:]-x[0:-1]>=0)
matplotlib.tests.test_axes.test_nonfinite_limits.test ... ok
I'd seen this on Sandro Tosi's output on this ML, is this a known numpy
issue? (using the Christoph Gohlke 1.6.2 MKL binaries)
Thanks,
RuiDC
-- 
View this message in context: http://old.nabble.com/v1.1.1rc2-tarballs-are-up-tp33987325p33993195.html
Sent from the matplotlib - devel mailing list archive at Nabble.com.

On Sat, Jun 9, 2012 at 11:14 PM, John Hunter <jd...@gm...> wrote:
> I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
>
> https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.1.1/
Debian package built and uploaded - thanks!
Cheers,
-- 
Sandro Tosi (aka morph, morpheus, matrixhasu)
My website: http://matrixhasu.altervista.org/
Me at Debian: http://wiki.debian.org/SandroTosi

Hi all,
I just sent in a pull request #941 to update the documentation for 
axes.arrow, because width,head_width,head_length, etc weren't 
documented, but I ran into a couple of concerns that I thought to discuss:
1. the default width of 0.001 seems absurdly small. 0.05 seems better. 
You can tell there's actually an arrowhead there with reasonable x- and 
y-limits.
2. I believe that length_includes_head ought to default to True. 
especially since the doc says: Draws arrow on specified axis from (*x*, 
*y*) to (*x* + *dx*, *y* + *dy*), one would expect by default the tip of 
the arrow to end at *x* + *dx*, *y* + *dy*
3. for the 'shape' arg, 'left' and 'right' seem to be backwards to me. 
Given that the arrow defines a vector direction, say \hat{x}, then 
'left' to me means only the part of the arrow in the +y half of the 
plane should be drawn.
M

Thanks -- I've already started the process for the files, but can you
test upload something into the 1.1 folder (easy to subsequently
delete). I just made you a project admin. I'd like to get this
figure out eventually!
On Sat, Jun 9, 2012 at 8:53 PM, Christoph Gohlke <cg...@uc...> wrote:
> Hi John,
>
> the matplotlib-1.1.1rc2.win installers are at
> <http://www.lfd.uci.edu/~gohlke/pythonlibs/#matplotlib>. I am again unable
> to upload them to SF.
>
> They are built against numpy 1.6.2 and include the test files. All tests
> pass on Python 2.7, 32 and 64 bit.
>
> Christoph
>
>
>
> On 6/9/2012 2:14 PM, John Hunter wrote:
>>
>> I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
>>
>>
>> https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.1.1/
>>
>> As soon as we get binaries, I'll send out another call for testing on
>> the users list. Russell and Christoph, easiest if you just upload the
>> binaries directly. You should both have permissions on the sf site.
>>
>> A copy of the site docs are available at
>> http://matplotlib.sourceforge.net/rc/v1.1.1rc2/
>>
>>
>

Hi John,
the matplotlib-1.1.1rc2.win installers are at 
<http://www.lfd.uci.edu/~gohlke/pythonlibs/#matplotlib>. I am again 
unable to upload them to SF.
They are built against numpy 1.6.2 and include the test files. All tests 
pass on Python 2.7, 32 and 64 bit.
Christoph
On 6/9/2012 2:14 PM, John Hunter wrote:
> I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
>
> https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.1.1/
>
> As soon as we get binaries, I'll send out another call for testing on
> the users list. Russell and Christoph, easiest if you just upload the
> binaries directly. You should both have permissions on the sf site.
>
> A copy of the site docs are available at
> http://matplotlib.sourceforge.net/rc/v1.1.1rc2/
>
>

Don't forget to tag the rc in git!
On Sat, Jun 9, 2012 at 6:40 PM, John Hunter <jd...@gm...> wrote:
> On Sat, Jun 9, 2012 at 5:33 PM, John Hunter <jd...@gm...> wrote:
> > On Sat, Jun 9, 2012 at 5:27 PM, Eric Firing <ef...@ha...> wrote:
> >> It also has backend_gdk.c (supposed to be a temporary copy of
> >> _backend_gdk.c) and some others like it.
> >
> > OK, I'll rebuild them super clean... I know what went wrong.
>
> Just uploaded two new files built from clean git checkouts.
>
>
> ------------------------------------------------------------------------------
> 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
>

On Sat, Jun 9, 2012 at 5:33 PM, John Hunter <jd...@gm...> wrote:
> On Sat, Jun 9, 2012 at 5:27 PM, Eric Firing <ef...@ha...> wrote:
>> It also has backend_gdk.c (supposed to be a temporary copy of
>> _backend_gdk.c) and some others like it.
>
> OK, I'll rebuild them super clean... I know what went wrong.
 Just uploaded two new files built from clean git checkouts.

On Sat, Jun 9, 2012 at 5:27 PM, Eric Firing <ef...@ha...> wrote:
> It also has backend_gdk.c (supposed to be a temporary copy of
> _backend_gdk.c) and some others like it.
OK, I'll rebuild them super clean... I know what went wrong.

On 06/09/2012 12:21 PM, John Hunter wrote:
> On Sat, Jun 9, 2012 at 5:11 PM, Eric Firing<ef...@ha...> wrote:
>> On 06/09/2012 11:14 AM, John Hunter wrote:
>>> I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
>>
>> John,
>>
>> Is it intentional that the tarballs include doc/build?
>
> It isn't intentional and it looks like it only affected the "notests"
> dist. Fixed it -- thanks for catching it.
It also has backend_gdk.c (supposed to be a temporary copy of 
_backend_gdk.c) and some others like it.
Which raises the question: why are these temporary copies needed at all? 
I don't see any explanation in setupext.py. I imagine it has to do with 
how some systems handle the leading underscore on the name, but it would 
be nice to have it explained in a comment in setupext.py.
Eric

On Sat, Jun 9, 2012 at 5:11 PM, Eric Firing <ef...@ha...> wrote:
> On 06/09/2012 11:14 AM, John Hunter wrote:
>> I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
>
> John,
>
> Is it intentional that the tarballs include doc/build?
It isn't intentional and it looks like it only affected the "notests"
dist. Fixed it -- thanks for catching it.

On 06/09/2012 11:14 AM, John Hunter wrote:
> I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
John,
Is it intentional that the tarballs include doc/build?
Eric
>
> https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.1.1/
>
> As soon as we get binaries, I'll send out another call for testing on
> the users list. Russell and Christoph, easiest if you just upload the
> binaries directly. You should both have permissions on the sf site.
>
> A copy of the site docs are available at
> http://matplotlib.sourceforge.net/rc/v1.1.1rc2/
>
> ------------------------------------------------------------------------------
> 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

I just uploaded the v1.1.1rc2 tarballs to the sourceforge site
https://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.1.1/
As soon as we get binaries, I'll send out another call for testing on
the users list. Russell and Christoph, easiest if you just upload the
binaries directly. You should both have permissions on the sf site.
A copy of the site docs are available at
http://matplotlib.sourceforge.net/rc/v1.1.1rc2/

Alan Griffiths <ala...@gm...> writes:
> I've come across a problem (possibly a bug in cleanup_path) after
> plotting a line containing nan's and then scaling the axes so that the
> line is clipped by the axes.
>
> Instead of a gap at the position of the nan value, two extra segments
> appear - back to the start and then onto the next-plus-one valid
> point. 
This sounds a lot like
https://github.com/matplotlib/matplotlib/issues/804
There was a bug in the fix for that, which was fixed in 
https://github.com/matplotlib/matplotlib/pull/873
I think at least the latter fix was merged later than the release
candidate was announced. You might want to try a later version.
-- 
Jouni K. Seppänen
http://www.iki.fi/jks