matplotlib-users Mailing List for matplotlib

In article 
<CAA...@ma...>,
 C M <cmp...@gm...> wrote:

> > > Trying to help a Mac friend running OSX 10.7 (Lion) easily set up to test
> > > scripts I send him, and have some questions:
> > >
> > > 1) Can Matplotlib 1.1 run on the Python 2.7.2 version that comes with
> > > Lion?
> >
> > Yes. You can easily build it yoursef as long as you have XCode
> > installed:
> >
> > - Edit setupext.py so that the list of values for "darwin" is
> > ['/usr/local', '/usr', /usr/X11']
> > (and for a really vanilla build leave out /usr/local). (I have a pull
> > request for this to be part of the standard distro, but I don't know if
> > or when it will go in since not a big issue.)
> >
> > Then do the usual:
> > % python setup.py build
> > % sudo python setup.py install
> 
> Where are you getting matplotlib 1.1 for Mac OSX 10.7 from to build it?

1.1.0 is the current release. I downloaded the unix tarball.

> > > 2) When is there expected to be an installer for Matplotlib 1.1 for OSX
> > > 10.7?
> >
> > There is one, but like all the matplotlib (and numpy and scipy) official
> > binaries it uses python.org's python, not Apple's.
> >
> > To go this route install python.org's Python 2.7.2 for 10.6-and-later
> > (which is 64-bit) and then install the official numpy and matpotlib
> > binaries.
> 
> >From where?  I didn't see it.  I am talking about having a binary
> installer on the Matplotlib downloads page,,. the last one I see there
> is this, which is for OSX-10.6:
> 
> matplotlib-1.1.0-py2.7-python.org-macosx10.6.dmg        2012-02-15

That is the one. 10.6 means "10.6 and later". It is build for 
python.org's Mac binary installer for 10.6 and later, which works just 
fine on Lion.

-- Russell

> > Trying to help a Mac friend running OSX 10.7 (Lion) easily set up to test
> > scripts I send him, and have some questions:
> >
> > 1) Can Matplotlib 1.1 run on the Python 2.7.2 version that comes with
> > Lion?
>
> Yes. You can easily build it yoursef as long as you have XCode
> installed:
>
> - Edit setupext.py so that the list of values for "darwin" is
> ['/usr/local', '/usr', /usr/X11']
> (and for a really vanilla build leave out /usr/local). (I have a pull
> request for this to be part of the standard distro, but I don't know if
> or when it will go in since not a big issue.)
>
> Then do the usual:
> % python setup.py build
> % sudo python setup.py install

Where are you getting matplotlib 1.1 for Mac OSX 10.7 from to build it?


> > 2) When is there expected to be an installer for Matplotlib 1.1 for OSX
> > 10.7?
>
> There is one, but like all the matplotlib (and numpy and scipy) official
> binaries it uses python.org's python, not Apple's.
>
> To go this route install python.org's Python 2.7.2 for 10.6-and-later
> (which is 64-bit) and then install the official numpy and matpotlib
> binaries.

>From where?  I didn't see it.  I am talking about having a binary
installer on the Matplotlib downloads page,,. the last one I see there
is this, which is for OSX-10.6:

matplotlib-1.1.0-py2.7-python.org-macosx10.6.dmg        2012-02-15

which is from about two month ago.

Thanks,
Che

All,

I often produce images for use in .KML with Google Earth.  Sometimes I need
the masking effect of matplotlib.toolkits.basemap.Basemap.fillcontinents
but in the case of Google Earth, I need to make continents transparent on
the final image.  Any ideas on how to do this?

Thanks!

Bruce

---------------------------------------
Bruce W. Ford

Hello

2012/4/24 Eric Firing <ef...@ha...>:
> On Tue, Apr 24, 2012 at 12:21 AM, Tony Yu <ts...@gm...> wrote:
>> On Mon, Apr 23, 2012 at 5:09 AM, jul tayon <jt...@gm...> wrote:
>>> Hello list
>>> Exected Result :
>>> https://github.com/jul/pypi-stat/blob/master/why.png
>>>
>> Sorry, I'm not clear on what the bug is. Is the picture you link to the
>> *expected* result or the result from *executing* the code snippet. There are
>> some inconsistencies between the pic and the code (legend, time on xaxis, 3
>> point in pic, title), so it's a bit unclear what you're referring to.
> With recent mpl you can use the ticklabel_format pyplot function or axes method.
>> If this isn't the bug you were suggesting, please clarify.

This was the feature I did not grasp.

Thanks for your help list. :)
Any chance there exists an HOWTO make nice looking graphs/plots for
tasteblind developpers   ?
Cheers,

On Tue, Apr 24, 2012 at 12:21 AM, Tony Yu <ts...@gm...> wrote:
>
>
> On Mon, Apr 23, 2012 at 5:09 AM, jul tayon <jt...@gm...> wrote:
>>
>> Hello list
>>
>> matplotlib website says this mailing  is the prefered way to report
>> bug, so here I am :
>>
>> System :
>> python ; 2.7.2+
>> matplotlib 1.1.0
>> ubuntu TLS amd64
>> matplotlib Backend : TkAgg
>>
>>
>> How to reproduce (on my pf)
>> Code Snippet :
>>
>> ######CODE
>>
>> import matplotlib
>> import matplotlib.pyplot as p
>> import random
>> y = random.randint(0,10000000)
>> p.plot( [ int( y * ( 1 - 0.0006)) , y])
>> p.show()
>>
>> ######ENDCODE
>>
>> Exected Result :
>> https://github.com/jul/pypi-stat/blob/master/why.png
>>
>> Cause ?
>>
>> It stinks something like a rounding error on the float.
>> If it also happens in Qt/Gtk backend it might be a matplotlib bug, if
>> it is only in Tk then it is either a Tk or a TkAgg bug.
>>
>> Cheers,
>> --
>> Jul
>
>
> Sorry, I'm not clear on what the bug is. Is the picture you link to the
> *expected* result or the result from *executing* the code snippet. There are
> some inconsistencies between the pic and the code (legend, time on xaxis, 3
> point in pic, title), so it's a bit unclear what you're referring to.
>
> If you're talking about the offset value added to the y-axis, it's actually
> not a bug; it's a feature (usually that's a joke, but in this case, I'm
> serious).
>
> This feature is useful when you have large numbers that vary over small
> ranges. Maybe it's not so bad in your example, but for larger numbers with
> smaller ranges (see code snippet below), it's quite helpful. I thought there
> was an rc parameter to tweak the value for using offsets, but I can't seem
> to find it. Below I create a custom tick formatter and turn offsets. There
> may be a better way to do this.

With recent mpl you can use the ticklabel_format pyplot function or axes method.

Eric

>
> If this isn't the bug you were suggesting, please clarify.
>
> Best,
> -Tony
>
> #~~~~~
> import matplotlib.pyplot as plt
>
> fig, ax = plt.subplots()
> # or `ax = plt.gca()` for older versions of matplotlib
>
> y = 1e9
> ax.plot([y, y+1])
> ax.yaxis.set_major_formatter(plt.ScalarFormatter(useOffset=False))
>
> plt.show()
> #~~~~~
>
> ------------------------------------------------------------------------------
> 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-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

On Mon, Apr 23, 2012 at 5:09 AM, jul tayon <jt...@gm...> wrote:

> Hello list
>
> matplotlib website says this mailing  is the prefered way to report
> bug, so here I am :
>
> System :
> python ; 2.7.2+
> matplotlib 1.1.0
> ubuntu TLS amd64
> matplotlib Backend : TkAgg
>
>
> How to reproduce (on my pf)
> Code Snippet :
>
> ######CODE
>
> import matplotlib
> import matplotlib.pyplot as p
> import random
> y = random.randint(0,10000000)
> p.plot( [ int( y * ( 1 - 0.0006)) , y])
> p.show()
>
> ######ENDCODE
>
> Exected Result :
> https://github.com/jul/pypi-stat/blob/master/why.png
>
> Cause ?
>
> It stinks something like a rounding error on the float.
> If it also happens in Qt/Gtk backend it might be a matplotlib bug, if
> it is only in Tk then it is either a Tk or a TkAgg bug.
>
> Cheers,
> --
> Jul
>

Sorry, I'm not clear on what the bug is. Is the picture you link to the
*expected* result or the result from *executing* the code snippet. There
are some inconsistencies between the pic and the code (legend, time on
xaxis, 3 point in pic, title), so it's a bit unclear what you're referring
to.

If you're talking about the offset value added to the y-axis, it's actually
not a bug; it's a feature (usually that's a joke, but in this case, I'm
serious).

This feature is useful when you have large numbers that vary over small
ranges. Maybe it's not so bad in your example, but for larger numbers with
smaller ranges (see code snippet below), it's quite helpful. I thought
there was an rc parameter to tweak the value for using offsets, but I can't
seem to find it. Below I create a custom tick formatter and turn offsets.
There may be a better way to do this.

If this isn't the bug you were suggesting, please clarify.

Best,
-Tony

#~~~~~
import matplotlib.pyplot as plt

fig, ax = plt.subplots()
# or `ax = plt.gca()` for older versions of matplotlib

y = 1e9
ax.plot([y, y+1])
ax.yaxis.set_major_formatter(plt.ScalarFormatter(useOffset=False))

plt.show()
#~~~~~