matplotlib-users Mailing List for matplotlib

Hello,

I'm in the process of migrating some software I've written from a
machine with MPL 0.86.2 on it to another with 0.87.2 .

The app is Qt-based with MPL embedded as a widget.  The plot itself uses
dates on the x-axis.

It appears that the behavior of autoscale_view() changed quite a bit
somewhere between these two versions.  

In my application I create a patches.Rectangle that is 100 seconds wide:

    start_time_num = matplotlib.dates.date2num( start_time )
    lower_left = [ matplotlib.dates.date2num( start_time ), start_y ]
    width = matplotlib.dates.date2num( start_time + datetime.timedelta( seconds=10 ) ) - start_time_num
    r = matplotlib.patches.Rectangle( lower_left, width, height )

With the 0.86.2 version, the total span of the x-axis after an
autoscale_view() call is about 12 seconds.  For the 0.87.2 version
however, the span is more than 200 years. (see attached .png files)

I don't see this type of behavior in the pylab environment, yet I'm
having a hard time creating a concise example that demonstrates the
behavior.  

I can keep working on extracting an snippet, but I thought I'd ask
if this is some intended behavioral change, and if so, if there is a
workaround.

Thank you!
Glen Mabey

On 4/19/06, Marin Manuel <mar...@gm...> wrote:
>
> christophe wrote:
>
> > Hi,
> >
> > Type error : RuntimeError*
> > *Error value : Could not open file /home/image.png
> >
> What path did you try ? From this error message it seems you tried to
> save in the /home folder where you dont have write permission. Try to
> save in /home/christophe/image.png (replace christophe by whatever your
> user name is)
>

it's the same problem with any path

It sounds like there are two questions here: whether the colorbar=20
correctly tracks the actual color mapping used in the image or contour=20
plot or whatever (it should always do this, but I think that at present=20
it does not), and whether the limits should change even if the mapping=20
does not.  I think the behavior you request is sensible, so that would=20
be my target, either as a standard behavior or as an option.  Thanks for=20
pointing out this aspect of the problem.

There is another problem with the colorbar code, and I have been trying=20
to decide how to handle that before looking at the update question, so I=20
have not resolved either one yet.

Eric

Jo=C3=A3o Lu=C3=ADs Silva wrote:
> Eric Firing wrote:
>=20
>> Andrea,
>>
>>  I agree, the colorbar should dynamically update.  A workaround might=20
>> be to simply call colorbar every time you update the image data.
>>
>> At least part of a longer-term solution may be to change colorbar so=20
>> that instead of always using contourf, it uses imshow when it is being=
=20
>> used with a nearly continuous set of colors instead of a small=20
>> discrete set.
>>
>> I have not had time to look at this closely, but I can add it to my li=
st.
>>
>> Eric
>>
>=20
> Although updating the colorbar when the image data changes should be th=
e=20
> default behaviour will there be a way to keep the colorbar from=20
> changing? I've got a visualization application where an image changes=20
> according to the position it's currently displaying but I want to keep=20
> the colorbar that reflects the largest value. Right now I just display=20
> first the image that has the largest value and the colorbar stays the=20
> same. Whatever the buggy behaviour an application has someone is taking=
=20
> advantage of it :-)
>=20
> Jo=C3=A3o Silva
>=20
>=20
>=20
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting lang=
uage
> that extends applications into web and mobile media. Attend the live=20
> webcast
> and join the prime developer group breaking into this new coding territ=
ory!
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D110944&bid=3D241720&dat=
=3D121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

How can one partially overlay text on top of text so as to obscure
the underlying text?

I want to print a beginning date at (0,0) and the current date
at (0.1,0) such that the current date's bounding box masks the
underlying beginning date.  What currently happens is the two dates
blend together to form a mess.  (I am creating an mpeg4 where the
current date marches from beginning to end.)

I have tried two things: set a bounding box, set a background color.

  rectangle = matplotlib.patches.Rectangle( (0,0), 0, 0 )
  rectangle.set_fill( True )

  # Drawing beginning date
  s = "%s UTC" % beginning.strftime( "%Y-%m-%d %H:%M" )
  t = matplotlib.text.Text( 0.0, 0.0, s )
  t.set_bbox( rectprops = rectangle.__dict__ )
  t.set_backgroundcolor( color='r' )
  axes.add_artist( t )

  # Draw current date
  txt = "%s UTC" % now.strftime( "%Y-%m-%d %H:%M" )
  t = matplotlib.text.Text( x, 0.0, s )
  t.set_bbox( rectprops = rectangle.__dict__ )
  t.set_backgroundcolor( color='r' )
  axes.add_artist( t )

The set_background color seems to do nothing.
The set_bbox raises the following--
  File "/usr/lib/python2.4/site-packages/matplotlib/artist.py", line 
193, in update
    raise AttributeError('Unknown property %s'%k)
AttributeError: Unknown property _facecolor

Thanks,
Andrew
Python 2.4, __version__  = '0.85', __revision__ = '$Revision: 1.83 $'

christophe wrote:

> Hi,
>
> Type error : RuntimeError*
> *Error value : Could not open file /home/image.png
>
What path did you try ? From this error message it seems you tried to 
save in the /home folder where you dont have write permission. Try to 
save in /home/christophe/image.png (replace christophe by whatever your 
user name is)

Here is a one-line patch to make setup complain if tk.h isn't found.

St=E9fan

Hi Etienne

On Wed, Apr 19, 2006 at 11:02:41AM -0400, Etienne Ringuet wrote:
> I am trying to build matplotlib on a freshly installed CentOS 4 worksta=
tion.
>=20
>
> error: Command "gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe
> -m32 -march=3Di386 -mtune=3Dpentium4 -D_GNU_SOURCE -fPIC -fPIC
> -I/usr/share/tcl8.4/../../include -I/usr/share/tk8.4/../../include
> -I/usr/local/include -I/usr/include -I. -Isrc -Iswig -Iagg23/include -I=
.
> -I/usr/local/include -I/usr/include -I.
> -I/usr/share/tcl8.4/../../include/freetype2
> -I/usr/share/tk8.4/../../include/freetype2
> -I/usr/local/include/freetype2 -I/usr/include/freetype2 -I./freetype2
> -Isrc/freetype2 -Iswig/freetype2 -Iagg23/include/freetype2 -I./freetype=
2
> -I/usr/local/include/freetype2 -I/usr/include/freetype2 -I./freetype2
> -I/usr/include/python2.3 -c src/_tkagg.cpp -o
> build/temp.linux-i686-2.3/src/_tkagg.o" failed with exit status 1
>
> I also have an error at the beginning that says I miss tk.h, my system =
is in
> french so I will not paste this part.

I just ran into this problem again myself, on an Ubuntu machine.  It
has been present for a while.

A quick fix is to edit 'setup.py' and change the line

BUILD_TKAGG        =3D 'auto'

to

BUILD_TKAGG =3D 0

Another solution is to install the Tk development headers.  Under
Ubuntu it is in the package tk8.4-dev.

This is probably something that should be caught by setup.py
before building though!

Regards
St=E9fan

On Wednesday, 19 April 2006 17:02, Etienne Ringuet wrote:
> [root@institut-c33109 matplotlib-0.87.2]# rpm -qa | grep pygtk
> pygtk2-2.4.0-1

Have you installed the "devel" version of the package (contains 
headers required for compilation)?

Johann

Hi,

This is my first post to the list so hi everyone!

I am trying to build matplotlib on a freshly installed CentOS 4 workstation.

I have read the FAQ about build problems with pygtk 2.4 is the problem fixed in my version?

[root@institut-c33109 matplotlib-0.87.2]# rpm -qa | grep pygtk
pygtk2-2.4.0-1

When I try python setup.py build it ends with:

error: Command "gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe
-m32 -march=i386 -mtune=pentium4 -D_GNU_SOURCE -fPIC -fPIC
-I/usr/share/tcl8.4/../../include -I/usr/share/tk8.4/../../include
-I/usr/local/include -I/usr/include -I. -Isrc -Iswig -Iagg23/include -I.
-I/usr/local/include -I/usr/include -I.
-I/usr/share/tcl8.4/../../include/freetype2
-I/usr/share/tk8.4/../../include/freetype2
-I/usr/local/include/freetype2 -I/usr/include/freetype2 -I./freetype2
-Isrc/freetype2 -Iswig/freetype2 -Iagg23/include/freetype2 -I./freetype2
-I/usr/local/include/freetype2 -I/usr/include/freetype2 -I./freetype2
-I/usr/include/python2.3 -c src/_tkagg.cpp -o
build/temp.linux-i686-2.3/src/_tkagg.o" failed with exit status 1

I also have an error at the beginning that says I miss tk.h, my system is in french so I will not paste this part.

Thanks for your help and time,

Etienne

Hi,

when I use savefig() to save the image, it saves it in the bin directory 
of the instance (../instance/bin/)
when I try to pass the path in the name (savefig('path/image.png')), I 
get an error :*

Type error : RuntimeError*
*Error value : Could not open file /home/image.png

*How can I set the save directory ?
thks

>>>>> "andrea" == andrea gavana@tin it <and...@ti...> writes:

    andrea> Hello John & NG, thank you very much for your answer. I
    andrea> had actually make it working using Line2D instances for
    andrea> every point in the plot (every well location has its own
    andrea> Line2D object associated). It works reasonably well, but I
    andrea> have some doubts about my implementation...  First of all,
    andrea> I don't know how actually *delete* a Line2D from the plot

ax.lines.remove(someline)

    andrea> (when the well should be hidden) and, moreover, when I
    andrea> have a large number of wells it takes quite a long time to
    andrea> create all the Line2D instances and plot them...  Do you
    andrea> think it will work faster/better if I use a
    andrea> RegularPolyCollection?

Yes, plotting a large number of Line2D instances will be significantly
slower than using a collection.  Give it a try!

JDH

Hello John & NG,

    thank you very much for your answer. I had 
actually make it working using Line2D instances for every point in the 
plot (every well location has its own Line2D object associated). It 
works reasonably well, but I have some doubts about my 
implementation...
First of all, I don't know how actually *delete* a 
Line2D from the plot (when the well should be hidden) and, moreover, 
when I have a large number of wells it takes quite a long time to 
create all the Line2D instances and plot them...
Do you think it will 
work faster/better if I use a RegularPolyCollection?

Thank you very 
much for your suggestions.

Andrea.

>----Original Message----
>From: 
jdh...@ac...
>Date: 18-Apr-2006 16:16
>To: 
"and...@ti..."<and...@ti...>
>Cc: <matplotlib-
us...@li...>
>Subj: Re: [Matplotlib-users] Single Point 
Stored Plot
>
>>>>>> "andrea" == andrea gavana@tin it 
<and...@ti...> writes:
>
>    andrea> Hello NG, I am facing a 
small problem for which I was
>    andrea> unable to find a solution. 
Basically, I have to make a
>    andrea> scatter plot of x, y 
coordinates (they represents oil
>    andrea> wells positions) on a 2D 
map, but I have some constraints:
>
>    andrea> 1) I would like to 
have each point as a separate "plot
>    andrea> object", because the 
user may decide to hide/show only a
>    andrea> particular well, or 
show/hide them all, or show/hide
>    andrea> whatever combination of 
wells.  2) At the moment when the
>    andrea> class that holds a 
single well position/status is created,
>    andrea> the plot window 
may not be there (because is the user that
>    andrea> decides when to 
open a new plot window).
>
>    andrea> I have tried something like:
>
>Hey Andrea, sorry for the delay.  This is a neat problem and one that
>collections were designed to solve.  You probably want to use a
>RegularPolyCollection rather than a PolyCollection (the latter is for
>arbitrary polygons and you must compute all the vertices yourself).
>Here is an example that allows you to add and delete markers -- it
>inserts them with random positions and colors.  You should be able to
>adapt it to your application quite easily.
>
>import random
>from 
matplotlib.colors import colorConverter
>from matplotlib.collections 
import RegularPolyCollection 
>import matplotlib.cm as cm
>from pylab 
import figure, show, nx
>
>fig = figure()
>ax = fig.add_subplot(111, 
xlim=(0,1), ylim=(0,1), autoscale_on=False)
>ax.set_title("Press 'a' to 
add a point, 'd' to delete one")
># a single point
>offsets = 
[(0.5,0.5)]
>facecolors = [cm.jet(0.5)]
>
>collection = 
RegularPolyCollection(
>    fig.dpi,
>    numsides=5, # a pentagon
>    
rotation=0,
>    sizes=(50,), 
>    facecolors = facecolors,
>    
edgecolors = (colorConverter.to_rgba('black'),),
>    linewidths = 
(1,),
>    offsets = offsets,
>    transOffset = ax.transData,
>    )
>
>ax.add_collection(collection)
>
>def onpress(event):
>    """
>    
press 'a' to add a random point from the collection, 'd' to delete one
>    """
>    if event.key=='a':
>        x,y = nx.mlab.rand(2)
>        color = cm.jet(nx.mlab.rand())
>        offsets.append((x,y))
>        facecolors.append(color)
>        fig.canvas.draw()
>    elif 
event.key=='d':
>        N = len(offsets)
>        if N>0:
>            
ind = random.randint(0,N-1)
>            offsets.pop(ind)
>            
facecolors.pop(ind)
>            fig.canvas.draw()
>
>fig.canvas.
mpl_connect('key_press_event', onpress)
>
>show()
>

Eric Firing wrote:
> Andrea,
> 
>  I agree, the colorbar should dynamically update.  A workaround might be 
> to simply call colorbar every time you update the image data.
> 
> At least part of a longer-term solution may be to change colorbar so 
> that instead of always using contourf, it uses imshow when it is being 
> used with a nearly continuous set of colors instead of a small discrete 
> set.
> 
> I have not had time to look at this closely, but I can add it to my list.
> 
> Eric
> 

Although updating the colorbar when the image data changes should be the 
default behaviour will there be a way to keep the colorbar from 
changing? I've got a visualization application where an image changes 
according to the position it's currently displaying but I want to keep 
the colorbar that reflects the largest value. Right now I just display 
first the image that has the largest value and the colorbar stays the 
same. Whatever the buggy behaviour an application has someone is taking 
advantage of it :-)

João Silva