matplotlib-users Mailing List for matplotlib

>>>>> "Jouni" =3D=3D Jouni K Sepp=E4nen <jk...@ik...> writes:

    Jouni> Hi, One thing that I really like about Matlab's handle
    Jouni> graphics system is that it is self-documenting: if I do
    Jouni> h=3Dplot(...), I can query the current values of all
    Jouni> properties with get:

This is a very useful feature in matplotlib.  Thanks for your patch.
I ended up implementing this functionality, but using a different
approach than your query methods.  Instead, I added formatted strings
to the doc strings of each setter that describe what that function
accepts, eg

    def set_linestyle(self, s):
        """
Set the linestyle of the line

ACCEPTS: [ '-' | '--' | '-.' | ':' | 'steps' | 'None' ]
        """

set now parses this information and uses it to display the
properties.  In the CVS version of matplotlib, you can get this
information, as in matlab, with

# report accepts info on all properties
set(line)

# report accepts info on linestyle property
set(line, 'linestyle')

# print the value of all properties
get(line)

# print the value of the linestyle property
get(line, 'linestyle')

Very nice!  See the files examples/set_and_get.py in CVS.  Here is
some sample output

1 >>> lines =3D plot([1,2,3])

2 >>> set(lines)
    alpha: float
    antialiased or aa: [True | False]
    clip_box: a matplotlib.transform.Bbox instance
    clip_on: [True | False]
    color or c: any matplotlib color - see help(colors)
    dashes: sequence of on/off ink in points
    data: (array xdata, array ydata)
    data_clipping: [True | False]
    figure: a matplotlib.figure.Figure instance
    label: any string
    linestyle or ls: [ '-' | '--' | '-.' | ':' | 'steps' | 'None' ]
    linewidth or lw: float
    lod: [True | False]
    marker: [ '+' | ',' | '.' | '1' | '2' | '3' | '4' | '<' | '>' |
    'D' | 'H' | '^' | '_' | 'd' | 'h' | 'o' | 'p' | 's' | 'v' | 'x' |
    '|' ]
    markeredgecolor or mec: any matplotlib color - see help(colors)
    markeredgewidth or mew: float
    markerfacecolor or mfc: any matplotlib color - see help(colors)
    markersize or ms: float
    transform: a matplotlib.transform transformation instance
    vertical_offset: DEPRECATED
    visible: [True | False]
    xclip: (xmin, xmax)
    xdata: array
    yclip: (ymin, ymax)
    ydata: array

3 >>> .

Thanks for the suggestion!

JDH

>>>>> "Transier," == Transier, Frederik <fre...@sa...> writes:

    Frederik> Hello, while I was trying to build the matplotlib 0.64
    Frederik> with python 2.4 on my win xp machine

    Frederik> a lot of errors occurred. Has anyone managed to install
    Frederik> it under these conditions?

    Frederik> Or is there any location I can download the binaries
    Frederik> for python 2.4 and win xp?

    Frederik> Any help would be appreciated.

Building on windows is a pain.  I'll try to include a python2.4rc1
installer with the next release, possibly next week.  If you need
something sooner, you'll have to post some more information to the
list.  Have you read the instructions for building win32 in
setupext.py, and downloaded the win32_static file from the matplotlib
web site that is pointed to in setupext?

JDH
 

>>>>> "Darrell" == Darrell Silver <da...@cl...> writes:

    Darrell> Hi, first time user, first time installer...

    Darrell> Is there a howto for matplotlib installation on redhat
    Darrell> 7.3/as2.1 & as3.0?

    Darrell> I'm trying to install matplotlib on these platforms and
    Darrell> am running into a seemingly endless list of dependancies
    Darrell> for python2.2, numarray:

    Darrell> just on the 7.3 side, - freetype >= 2.1.7 is needed, but
    Darrell> rh has 2.0.9-2.  - pygtk >= 1.99.16 needed, rh has
    Darrell> 1.99.8-7.  - pygtk 2.2.0 needs glib >= 2.2.0, redhat has
    Darrell> 1.2.10-5, - pygtk 1.99.16 compiles, but I get error
    Darrell> "ImportError: /usr/lib/libpangoxft-1.0.so.0 undefinded
    Darrell> symbol: FT_Seek_Stream", which I haven't tracked down
    Darrell> just yet...  - the line "inf = infty = Infinity =
    Darrell> _ieee.inf" in na_imports.py fails to find 'inf' in
    Darrell> '_ieee', but commenting it out works fine.  - While the
    Darrell> install webpage says I only need freetype,libpng,zlib to
    Darrell> compile, I get runtime complaints on import of
    Darrell> matplotlib.matlab * (running simple_plot.py) about
    Darrell> missing image libraries if I don't turn the BUILD_IMAGE
    Darrell> flag on.

I have compiled matplotlib on 7.3 but it's been a while.  Rather than
upgrading to the latest pygtk, which will require a major upgrade of
all your gtk libs and their dependencies, try getting pygtk-1.99.16.
Since you already have pygtk-1.99.x, I think you should be able to
compile this.  And install freetype 2.1.7 or later, and numarray 1.1
or later, and try again.  With luck, with these packages you'll
get much farther.  

I suggest you try to compile pygtk with --enable-thread and
--enable-numpy

When building matplotlib, set the following in setup.py

BUILD_IMAGE     = 1
BUILD_AGG       = 1
BUILD_GTKAGG    = 1
BUILD_TKAGG     = 0
BUILD_WINDOWING = 0

If you can't build, please post the build output to the list and we'll
take it from there.

Good luck!
JDH


    Darrell> This is all way too difficult me as I'm not looking
    Darrell> forward to upgrading so many packages across our
    Darrell> production machines...am I missing something obvious?
    Darrell> Perhaps an older version of matplotlib known to work on
    Darrell> this redhat?  I'm just using it for simple time-series
    Darrell> plots, real-time and to some file (png, gif, ps, pdf...).


    Darrell> thanks in advance for any help,

    Darrell> Darrell



    Darrell> -------------------------------------------------------
    Darrell> SF email is sponsored by - The IT Product Guide Read
    Darrell> honest & candid reviews on hundreds of IT Products from
    Darrell> real users.  Discover which products truly live up to the
    Darrell> hype. Start reading
    Darrell> now. http://productguide.itmanagersjournal.com/
    Darrell> _______________________________________________
    Darrell> Matplotlib-users mailing list
    Darrell> Mat...@li...
    Darrell> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

>>>>> "christophe" == christophe grimault <chr...@no...> writes:


    christophe> The import of the numeric version of the _transforms
    christophe> module, _nc_transforms, failed.  This is either
    christophe> because numeric was unavailable when matplotlib was
    christophe> compiled, or because a dependency of _nc_transforms
    christophe> could not be satisfied.  If it appears that
    christophe> _nc_transforms was not built, make sure you have a
    christophe> working copy of numeric and then re-install
    christophe> matplotlib. Otherwise, the following traceback gives
    christophe> more details:

Try rm -rf ing your build directory and your site-packages/matplotlib
directory and do a clean rebuild and reinstall.  We've seen a lot of
problems not too different from yours when installing recent versions
of matplotlib over older versions.  If you're still encountering
troubles, post back and we'll try something different.

    christophe> ImportError:
    christophe> /usr/local/lib/python2.3/site-packages/matplotlib/_nc_transforms.so:
    christophe> undefined symbol: _ZNSs4_Rep20_S_empty_rep_storageE

This is a linker error, I don't think changing your include paths will
help.

    christophe> I tried many things in the setupext.py. Freetype is
    christophe> correctly installed and Numeric is in
    christophe> /opt/Numeric23.1. I added this to the
    christophe> script.... because the way it appends include/lib path
    christophe> seems a little buggy to me.

    christophe> I also tried other things. I still get the same error
    christophe> message. I'm stuck ! Has anyone encoutered this
    christophe> problem ?  solved it ? Shall i install a newer version
    christophe> of Numeric ? In an other place than /opt ?

matplotlib certainly works with Numeric 23.1.  I would advise you to
upgrade to the latest Numeric because many bugs have been fixed since
23.1, but it won't stop matplotlib from working.  

JDH

>>>>> "Daniel" == Daniel Newton <da...@ne...> writes:

    Daniel> Hi, I am trying to make a legend transparent (using
    Daniel> GTKAgg) as sometimes it is quite big an obscures part of
    Daniel> my graphs.

    Daniel> I have tryed this with no luck: l = figure.legend(lines,
    Daniel> names, legend) l.set_alpha(.5)

    Daniel> is this possible?

Is it the entire legend you want to make transparent, or simply the
rectangular background?  For the latter, you need to set the alpha on
the legend "frame".  You have two options here

leg = legend(blah, blah)
leg.draw_frame(False)  # turn it off entirely

or 

frame = leg.get_frame()
frame.set_alpha(0.5)   # make it semi-transparent

See http://matplotlib.sf.net/examples/legend_demo.py for an example of
getting handles to all the lines, patches and texts in the legend,
whose properties you can set independently.

There has been a discussion on whether it would be desirable for a set
call on an object to propagate to all the objects it contains.  Ie,
would it be a good thing to be able to do

set(leg, alpha=0.5) 

and have this propogate to the legend frame, lones, patches, and text
instances.  Or is the current setup where you control each of these
objects independently preferable?

JDH

Hi all,

I've successfully used version 0.52. Yesterday I installed v0.64 on my 
RH9 linux box. The build and install went OK. However, when I try to 
play with it, I get:

The import of the numeric version of the _transforms module, 
_nc_transforms, failed.
This is either because numeric was unavailable when matplotlib was compiled,
or because a dependency of _nc_transforms could not be satisfied.
If it appears that _nc_transforms was not built,  make sure you have a 
working copy of
numeric and then re-install matplotlib. Otherwise, the following 
traceback gives more details:

Traceback (most recent call last):
....
    from matplotlib.backends.backend_wxagg import FigureCanvasWxAgg as 
FigureCanvas
  File 
"/usr/local/lib/python2.3/site-packages/matplotlib/backends/__init__.py", 
line 20, in ?
    globals(),locals(),[backend_name])
  File 
"/usr/local/lib/python2.3/site-packages/matplotlib/backends/backend_wxagg.py", 
line 18, in ?
    from backend_agg import FigureCanvasAgg
  File 
"/usr/local/lib/python2.3/site-packages/matplotlib/backends/backend_agg.py", 
line 78, in ?
    from matplotlib.backend_bases import RendererBase,\
  File 
"/usr/local/lib/python2.3/site-packages/matplotlib/backend_bases.py", 
line 13, in ?
    from patches import Rectangle
  File "/usr/local/lib/python2.3/site-packages/matplotlib/patches.py", 
line 5, in ?
    from artist import Artist
  File "/usr/local/lib/python2.3/site-packages/matplotlib/artist.py", 
line 4, in ?
    from transforms import identity_transform
  File 
"/usr/local/lib/python2.3/site-packages/matplotlib/transforms.py", line 
188, in ?
    from _transforms import Value, Point, Interval, Bbox, Affine
  File 
"/usr/local/lib/python2.3/site-packages/matplotlib/_transforms.py", line 
11, in ?
    from matplotlib._nc_transforms import *
ImportError: 
/usr/local/lib/python2.3/site-packages/matplotlib/_nc_transforms.so: 
undefined symbol: _ZNSs4_Rep20_S_empty_rep_storageE

I tried many things in the setupext.py. Freetype is correctly installed 
and Numeric is in /opt/Numeric23.1. I added this to the script.... 
because the way it appends include/lib path seems a little buggy to me.

def add_ft2font_flags(module):
    'Add the module flags to build extensions which use gd'
    ....
    module.include_dirs.append('/usr/local/include/freetype2/freetype')
    ....
and
def add_agg_flags(module):
    'Add the module flags to build extensions which use agg'

    # before adding the freetype flags since -z comes later
    module.libraries.append('png')
    module.libraries.append('z')
    add_base_flags(module)
    module.include_dirs.extend(['src','agg22/include', '.'])
    module.include_dirs.append('/opt/Numeric-23.1/Include')
    # put these later for correct link order
    module.libraries.extend(['stdc++', 'm'])

I also tried other things. I still get the same error message. I'm stuck 
! Has anyone encoutered this problem ?
solved it ? Shall i install a newer version of Numeric ? In an other 
place than /opt ?

Any help would be geatly appreciated !

Christophe

Hi,

I am trying to make a legend transparent (using GTKAgg) as sometimes it 
is quite big an obscures part of my graphs.

I have tryed this with no luck:
        l = figure.legend(lines, names, legend)
        l.set_alpha(.5)

is this possible?

Thanks
Dan