matplotlib-users Mailing List for matplotlib (Page 3)

Hi,

your build worked out of the box on my macbook pro. I have the following
setup:

   - Python Universal Build from Python.org
   - Homebuld version of numpy 0.96
   - ipython 0.7.1

Thanks for the great work!

Markus


On 4/26/06, Charlie Moad <cw...@gm...> wrote:
>
> Hey all (primarily Chris),
>      I took the time to make a universal build that statically links
> universal libpng and freetype builds.  It was built against the latest
> pyarrays.  I don't know if anyone else got around to this yet.  It
> would be great if anyone out there with an intel mac could test it
> out.  If everything seems to work I'll send it to the pythonmac-sig
> group to be included in their package listing.
>
> http://euclid.uits.iupui.edu/~cmoad/mpleggs/
>
> Thanks,
>      Charlie
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronim=
o
> http://sel.as-us.falkag.net/sel?cmdlnk&kid=120709&bid&3057&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Qing Zhang wrote:
> Hi,
> 
> When I plot a contourf(X, Y, Z, V) map with a list of specified values V 
> for the contour lines, the map will have blank spaces where  Z are not 
> in the range specified by V. Is it possible to fill the blank spaces 
> with the two colors for the largest and the smallest V values?
> 
> Thanks,
> 
> Qing
> 

Qing,

Yes. One easy way to do this is is to put very large negative and 
positive values at the ends of your V array, for example

V = [-100, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 100]

It doesn't matter how large these outside values are, so long as they 
encompass all your data.  With the default contourf kwarg 
clip_ends=True, the color scaling will not use them.

Eric

Hi,

When I plot a contourf(X, Y, Z, V) map with a list of specified values V 
for the contour lines, the map will have blank spaces where  Z are not 
in the range specified by V. Is it possible to fill the blank spaces 
with the two colors for the largest and the smallest V values?

Thanks,

Qing

-- 
----------------------------------------------------------------
Qing Zhang, Ph.D.               Molecular Graphics Laboratory
Research Associate              Department of Molecular Biology
                                The Scripps Research Institute
Tel: 858-784-2333               10550 North Torrey Pines Road
Fax: 858-784-2860               Mail Drop: MB-5
Email: qz...@sc...       La Jolla, CA 92037
---------------------------------------------------------------- 

Hey all (primarily Chris),
     I took the time to make a universal build that statically links
universal libpng and freetype builds.  It was built against the latest
pyarrays.  I don't know if anyone else got around to this yet.  It
would be great if anyone out there with an intel mac could test it
out.  If everything seems to work I'll send it to the pythonmac-sig
group to be included in their package listing.

http://euclid.uits.iupui.edu/~cmoad/mpleggs/

Thanks,
     Charlie

On Tuesday 25 April 2006 17:27, Nadia Dencheva wrote:
> >  >>> tk.getvar('tk_library')
> >
> > u'/usr/stsci/pyssgdev/lib/tk8.4'
> >
> > (The first thing that comes to mind is 'unicode':))
>
> Hmm, turns out this is not a joke.
>
> On a linux machine I get
>
> enchanter> python
> Python 2.4.3 (#2, Apr 17 2006, 10:36:38)
> [GCC 3.2.3 20030502 (Red Hat Linux 3.2.3-54)] on linux2
> Type "help", "copyright", "credits" or "license" for more information.
>
>  >>> import Tkinter
>  >>> tk=Tkinter.Tk()
>  >>> type(tk.getvar('tk_library'))
>
> <type 'str'>
>
>  >>> type(tk.getvar('tcl_library'))
>
> <type 'str'>
>
> The same example on MacOSX (python, tcl/tk are X11 based, built in
>
> a local directory) gives:
>  >>> type(tk.getvar('tk_library'))
>
> <type 'unicode'>
>
>  >>> type(tk.getvar('tcl_library'))
>
> <type 'str'>
>
>
> I'm using tcl/tk 8.3 on linux and tcl/tk 8.4 on the mac.
> Is anyone aware of any changes to Tk 8.4 that could cause this
> behavior?
>
> A possible workaround is
>
> tk.getvar('tk_library').encode('utf8')

This might be related to a unicode bug in python too. I saw a mention 
somewhere that python-2.4.3 would include some unicode bug fixes.

> 
>  >>> tk.getvar('tk_library')
> u'/usr/stsci/pyssgdev/lib/tk8.4'

> (The first thing that comes to mind is 'unicode':))
> 
Hmm, turns out this is not a joke.

On a linux machine I get

enchanter> python
Python 2.4.3 (#2, Apr 17 2006, 10:36:38)
[GCC 3.2.3 20030502 (Red Hat Linux 3.2.3-54)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
 >>> import Tkinter
 >>> tk=Tkinter.Tk()
 >>> type(tk.getvar('tk_library'))
<type 'str'>
 >>> type(tk.getvar('tcl_library'))
<type 'str'>

The same example on MacOSX (python, tcl/tk are X11 based, built in
a local directory) gives:

 >>> type(tk.getvar('tk_library'))
<type 'unicode'>
 >>> type(tk.getvar('tcl_library'))
<type 'str'>


I'm using tcl/tk 8.3 on linux and tcl/tk 8.4 on the mac.
Is anyone aware of any changes to Tk 8.4 that could cause this
behavior?

A possible workaround is

tk.getvar('tk_library').encode('utf8')

Nadia

Jim,



James Boyle wrote:
> Thanks for the replies - since it worked Ok for other people I looked a 
> bit further:
> 
> matplotlib 0.87.2,   OS X  10.3.9,    Numeric - 23.7,  ipython 0.6.15
> 
>  running quadmesh_demo.py in examples gives a bus error when the figure 
> is rendered.
> 
> It turns out that the collections object returned by the line 16 of 
> quadmesh_demo.py:
> ax.pcolormesh(Qx,Qz,Z)
> 
> returns vertices with complex values i.e.
> b = ax.pcolormesh(Qx,Qz,Z)
> c= b._coordinates
> c[0]
> Out[23]: [ 1.10000002+0.j,-1.99498999+0.j,]
> 
> this happens right away in pcolormesh since _coordinates are set there.
> If I  force a type cast  as in:
> b._coodinates = b._coodinates.astype(Numeric.Float32)
> all goes well
> 
>  I suspected that something was screwy in lines of pcolormesh (lines 
> 2422 - 2326) in axes.py:
>         coords = zeros(((Nx * Ny), 2),"Float32")

In svn I took the quotes out; if I understand correctly, this should 
solve the problem.  Correct?

Eric



>         # Numeric and numpy refuse to cast the Float64 arrays
>         # to Float32 with simple assignment, so we do it explicitly.
>         coords[:, 0] = X.astype(Float32)
>         coords[:, 1] = Y.astype(Float32)
> 
>  it appears to be OK for a Float32 BUT if there is a string 
> representation "Float32'"  the Numeric module  takes only the first 
> character in this case an 'F' which implies complex for Numeric.
> all appears OK for numarray and numpy.
> 
> ---Jim
> 
> 
> 
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job 
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On Apr 25, 2006, at 6:59 AM, <kei...@bt...> wrote:

> After "from pylab import *", log(0.0) gives
>
> Warning: Encountered divide by zero(s)  in log
>
> Where is this behaviour documented?   What does it mean?   Why  
> should log(0.0) divide by zero?   What does the plural "zeros"  
> refer to?   Why break the IEEE standard that log(0.0) should  
> silently return -Inf?

Adding to Robert's reply, I don't believe that the standard mandates  
that programs should silently return Inf in the presence of divide by  
zero errors. It does allow for  programs to determine how they want  
to respond to such errors, and numpy allows you to  customize that.  
You can ignore them, print a warning (the default I believe), or  
raise an exception. Believe it or not, not everyone wants the same  
behavior for all cases.

Perry

kei...@bt... wrote:
> After "from pylab import *", log(0.0) gives
> 
> Warning: Encountered divide by zero(s)  in log
> 
> Where is this behaviour documented?   What does it mean?   Why should log(0.0) divide by zero?   What does the plural "zeros" refer to?   Why break the IEEE standard that log(0.0) should silently return -Inf?

http://www.opengroup.org/onlinepubs/000095399/functions/log.html

Look at the section "ERRORS", specifically "Pole Error". By and large the
implementation of log() comes from the underlying libm. The plural is because
log() is a ufunc and can take arrays, possibly with multiple zeros.

-- 
Robert Kern
rob...@gm...

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco

John Hunter wrote:
>>>>>>     Humufr> In [5]: tk.getvar('tk_library') Out[5]:
>>>>>>     Humufr> u'/usr/share/tcl/tk8.4'
>>>>>>     >>  This looks like what Eric is getting (offlist).
>>>>>>     >> 
>>>>>>     >> Can you confirm your tcl.h and tk.h are in /usr/include?
>>>>>>
>>>>>>     Humufr> confirm
>>>>>>
>>>>>> Where are your libtcl and libtk files?
>>>>>>
>>>>>>             
torro[~/tmp/firefox] ls /usr/lib/libtcl
libtcl8.4.so*     libtclstub8.4.a   libtclx8.4.so*   
torro[~/tmp/firefox] ls /usr/lib/libtk
libtk8.4.so*     libtkstub8.4.a  

The function is not working. I print the test and another Suse mystery 
appeared...

python setup.py install
False /usr/share/tcl/tcl8.4/../../include
Traceback (most recent call last):
  File "setup.py", line 229, in ?
    build_tkagg(ext_modules, packages, NUMERIX)
  File "/home/humufr/matplotlib/matplotlib/setupext.py", line 570, in 
build_tkagg
    add_tk_flags(module) # do this first
  File "/home/humufr/matplotlib/matplotlib/setupext.py", line 495, in 
add_tk_flags
    module.include_dirs.extend([o.tcl_inc, o.tk_inc])
AttributeError: 'NoneType' object has no attribute 'tcl_inc'

The result of the test to contourn Suse problem:

print os.path.exists(o.tcl_inc),o.tcl_inc

False /usr/share/tcl/tcl8.4/../../include

And I checked the directory:

torro[~/tmp/firefox] ls /usr/share/tcl/tcl8.4/../../include
ls: /usr/share/tcl/tcl8.4/../../include: No such file or directory


N.

>>>>>> Please try the attached, which isn't pretty but not much more we
>>>>>> can do on a broken system
>>>>>>
>>>>>>
>>>>>> def find_tcltk():
>>>>>>     """Finds Tcl/Tk includes/libraries/version by interrogating Tkinter."""
>>>>>>     try:
>>>>>>         import Tkinter
>>>>>>     except:
>>>>>>         print "Tkinter not properly installed\n"
>>>>>>         sys.exit(1)
>>>>>>     if Tkinter.TkVersion < 8.3:
>>>>>>         print "Tcl/Tk v8.3 or later required\n"
>>>>>>         sys.exit(1)
>>>>>>     o = FoundTclTk()
>>>>>>     try:
>>>>>>         tk=Tkinter.Tk()	
>>>>>>     except Tkinter.TclError:
>>>>>>         print "Using default library and include directories for Tcl and Tk because a"
>>>>>>         print "Tk window failed to open.  You may need to define DISPLAY for Tk to work"
>>>>>>         print "so that setup can determine where your libraries are located."
>>>>>>         o.tcl_lib = "/usr/local/lib"
>>>>>>         o.tcl_inc = "/usr/local/include"
>>>>>>         o.tk_lib = "/usr/local/lib"
>>>>>>         o.tk_inc = "/usr/local/include"
>>>>>>         o.tkv = ""
>>>>>>     else:
>>>>>>         tk.withdraw()
>>>>>>         o.tcl_lib = os.path.join((tk.getvar('tcl_library')), '../')
>>>>>>         o.tk_lib = os.path.join(str(tk.getvar('tk_library')), '../')
>>>>>>         o.tkv = str(Tkinter.TkVersion)[:3]
>>>>>>         o.tcl_inc = os.path.join((tk.getvar('tcl_library')), 
>>>>>>                     '../../include/tcl'+o.tkv)
>>>>>>         if not os.path.exists(o.tcl_inc):
>>>>>>             o.tcl_inc = os.path.join((tk.getvar('tcl_library')), 
>>>>>>                         '../../include')
>>>>>>         o.tk_inc = os.path.join((tk.getvar('tk_library')), 
>>>>>>                     '../../include/tk'+o.tkv)        
>>>>>>         if not os.path.exists(o.tk_inc):
>>>>>>             o.tk_inc = os.path.join((tk.getvar('tk_library')), 
>>>>>>                         '../../include')
>>>>>>
>>>>>>         # this is a hack for suse, which is broken
>>>>>>         if not os.path.exists(o.tcl_inc):            
>>>>>>             if os.path.exists('/usr/include/tcl.h') and os.path.exists('/usr/include/tk.h'):
>>>>>>                 o.tcl_inc = '/usr/include/'
>>>>>>                 o.tk_inc = '/usr/include/'
>>>>>>                 
>>>>>>         if not os.path.exists(o.tcl_inc):
>>>>>>             print 'cannot find tcl/tk headers. giving up.'
>>>>>>             sys.exit()
>>>>>>
>>>>>>
>>>>>>
>>>>>> -------------------------------------------------------
>>>>>> Using Tomcat but need to do more? Need to support web services, security?
>>>>>> Get stuff done quickly with pre-integrated technology to make your job easier
>>>>>> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>>>>>> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>>>>>> _______________________________________________
>>>>>> Matplotlib-users mailing list
>>>>>> Mat...@li...
>>>>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>>>>>
>>>>>>             

John Hunter wrote:
>>>>>>"Humufr" == Humufr  <hu...@ya...> writes:
> 
> 
>     Humufr> I looked a little bit at this problem before (I have the
>     Humufr> same issue) and I saw that the test on suse 10 (at least)
>     Humufr> give a bad answer and so matplotlib can't find the
>     Humufr> header. This is the result of the test on the suse install
>     Humufr> in my lab:
> 
>     Humufr> In [1]: import Tkinter
> 
>     Humufr> In [2]: tk=Tkinter.Tk()
> 
>     Humufr> In [3]: tk.getvar('tcl_library') Out[3]:
>     Humufr> '/usr/share/tcl/tcl8.4'
> 
>     Humufr> In [4]: tk.getvar('tk_library') Out[4]:
>     Humufr> u'/usr/share/tcl/tk8.4'
> 
>     Humufr> In [5]: tk.getvar('tk_library') Out[5]:
>     Humufr> u'/usr/share/tcl/tk8.4'


> It looks like the Suse tcl/tl/Tkinter install is broken -- is this an
> official package?  We should check the effbot's latest code (which is
> where this is borrowed from) to see if he has a fix in for this.
> 
> JDH

This may not be such a rare event.
I see the same problem on MacOSX 10.3: python, tcl/tk
(X11 based) compiled in a local directory (and is one of the reasons I don't
use the original setupext on MacOSX).


 >>> tk.getvar('tcl_library')
'/usr/stsci/pyssgdev/lib/tcl8.4'

but

 >>> tk.getvar('tk_library')
u'/usr/stsci/pyssgdev/lib/tk8.4'

I haven't had time to try to understand it but I believe the installation
I have is correct and this is a python/Tkinter problem.
(The first thing that comes to mind is 'unicode':))

Nadia

>>>>> "Humufr" == Humufr  <hu...@ya...> writes:

    Humufr> In [5]: tk.getvar('tk_library') Out[5]:
    Humufr> u'/usr/share/tcl/tk8.4'
    >>  This looks like what Eric is getting (offlist).
    >> 
    >> Can you confirm your tcl.h and tk.h are in /usr/include?

    Humufr> confirm

Where are your libtcl and libtk files?

Please try the attached, which isn't pretty but not much more we
can do on a broken system


def find_tcltk():
    """Finds Tcl/Tk includes/libraries/version by interrogating Tkinter."""
    try:
        import Tkinter
    except:
        print "Tkinter not properly installed\n"
        sys.exit(1)
    if Tkinter.TkVersion < 8.3:
        print "Tcl/Tk v8.3 or later required\n"
        sys.exit(1)
    o = FoundTclTk()
    try:
        tk=Tkinter.Tk()	
    except Tkinter.TclError:
        print "Using default library and include directories for Tcl and Tk because a"
        print "Tk window failed to open.  You may need to define DISPLAY for Tk to work"
        print "so that setup can determine where your libraries are located."
        o.tcl_lib = "/usr/local/lib"
        o.tcl_inc = "/usr/local/include"
        o.tk_lib = "/usr/local/lib"
        o.tk_inc = "/usr/local/include"
        o.tkv = ""
    else:
        tk.withdraw()
        o.tcl_lib = os.path.join((tk.getvar('tcl_library')), '../')
        o.tk_lib = os.path.join(str(tk.getvar('tk_library')), '../')
        o.tkv = str(Tkinter.TkVersion)[:3]
        o.tcl_inc = os.path.join((tk.getvar('tcl_library')), 
                    '../../include/tcl'+o.tkv)
        if not os.path.exists(o.tcl_inc):
            o.tcl_inc = os.path.join((tk.getvar('tcl_library')), 
                        '../../include')
        o.tk_inc = os.path.join((tk.getvar('tk_library')), 
                    '../../include/tk'+o.tkv)        
        if not os.path.exists(o.tk_inc):
            o.tk_inc = os.path.join((tk.getvar('tk_library')), 
                        '../../include')

        # this is a hack for suse, which is broken
        if not os.path.exists(o.tcl_inc):            
            if os.path.exists('/usr/include/tcl.h') and os.path.exists('/usr/include/tk.h'):
                o.tcl_inc = '/usr/include/'
                o.tk_inc = '/usr/include/'
                
        if not os.path.exists(o.tcl_inc):
            print 'cannot find tcl/tk headers. giving up.'
            sys.exit()

John Hunter wrote:
>
>     Humufr> I looked a little bit at this problem before (I have the
>     Humufr> same issue) and I saw that the test on suse 10 (at least)
>     Humufr> give a bad answer and so matplotlib can't find the
>     Humufr> header. This is the result of the test on the suse install
>     Humufr> in my lab:
>
>     Humufr> In [1]: import Tkinter
>
>     Humufr> In [2]: tk=Tkinter.Tk()
>
>     Humufr> In [3]: tk.getvar('tcl_library') Out[3]:
>     Humufr> '/usr/share/tcl/tcl8.4'
>
>     Humufr> In [4]: tk.getvar('tk_library') Out[4]:
>     Humufr> u'/usr/share/tcl/tk8.4'
>
>     Humufr> In [5]: tk.getvar('tk_library') Out[5]:
>     Humufr> u'/usr/share/tcl/tk8.4'
>
> This looks like what Eric is getting (offlist).
>
> Can you confirm your tcl.h and tk.h are in /usr/include?  

confirm

> Do the dirs
> /usr/include/tcl8.4 and /usr/include/tk8.4 exist?
>
>   
no they doesn't exist. All the header for tcl/tk are install in 
/usr/include

> It looks like the Suse tcl/tl/Tkinter install is broken -- is this an
> official package?  We should check the effbot's latest code (which is
> where this is borrowed from) to see if he has a fix in for this.
>
>   
I think so but like  I told before it's not the only package a little 
bit strange on Suse...

N.

> JDH
>
>   

Thanks for the replies - since it worked Ok for other people I looked a 
bit further:

matplotlib 0.87.2,   OS X  10.3.9,    Numeric - 23.7,  ipython 0.6.15

  running quadmesh_demo.py in examples gives a bus error when the figure 
is rendered.

It turns out that the collections object returned by the line 16 of 
quadmesh_demo.py:
ax.pcolormesh(Qx,Qz,Z)

returns vertices with complex values i.e.
b = ax.pcolormesh(Qx,Qz,Z)
c= b._coordinates
c[0]
Out[23]: [ 1.10000002+0.j,-1.99498999+0.j,]

this happens right away in pcolormesh since _coordinates are set there.
If I  force a type cast  as in:
b._coodinates = b._coodinates.astype(Numeric.Float32)
all goes well

  I suspected that something was screwy in lines of pcolormesh (lines 
2422 - 2326) in axes.py:
         coords = zeros(((Nx * Ny), 2),"Float32")
         # Numeric and numpy refuse to cast the Float64 arrays
         # to Float32 with simple assignment, so we do it explicitly.
         coords[:, 0] = X.astype(Float32)
         coords[:, 1] = Y.astype(Float32)

  it appears to be OK for a Float32 BUT if there is a string 
representation "Float32'"  the Numeric module  takes only the first 
character in this case an 'F' which implies complex for Numeric.
all appears OK for numarray and numpy.

---Jim

>>>>> "Humufr" == Humufr  <hu...@ya...> writes:

    Humufr> I looked a little bit at this problem before (I have the
    Humufr> same issue) and I saw that the test on suse 10 (at least)
    Humufr> give a bad answer and so matplotlib can't find the
    Humufr> header. This is the result of the test on the suse install
    Humufr> in my lab:

    Humufr> In [1]: import Tkinter

    Humufr> In [2]: tk=Tkinter.Tk()

    Humufr> In [3]: tk.getvar('tcl_library') Out[3]:
    Humufr> '/usr/share/tcl/tcl8.4'

    Humufr> In [4]: tk.getvar('tk_library') Out[4]:
    Humufr> u'/usr/share/tcl/tk8.4'

    Humufr> In [5]: tk.getvar('tk_library') Out[5]:
    Humufr> u'/usr/share/tcl/tk8.4'

This looks like what Eric is getting (offlist).

Can you confirm your tcl.h and tk.h are in /usr/include?  Do the dirs
/usr/include/tcl8.4 and /usr/include/tk8.4 exist?

It looks like the Suse tcl/tl/Tkinter install is broken -- is this an
official package?  We should check the effbot's latest code (which is
where this is borrowed from) to see if he has a fix in for this.

JDH

I looked a little bit at this problem before (I have the same issue) and 
I saw that the test on suse 10 (at least) give a bad answer and so 
matplotlib can't find the header. This is the result of the test on the 
suse install in my lab:

In [1]: import Tkinter

In [2]: tk=Tkinter.Tk()

In [3]: tk.getvar('tcl_library')
Out[3]: '/usr/share/tcl/tcl8.4'

In [4]: tk.getvar('tk_library')
Out[4]: u'/usr/share/tcl/tk8.4'

In [5]: tk.getvar('tk_library')
Out[5]: u'/usr/share/tcl/tk8.4'

In [6]: Tkinter.TkVersion
Out[6]: 8.4000000000000004

That explain the problem with setupext.py and like this problem came for 
the python install on a specific distribution I don't see how that can 
be generalized. I didn't have this kind of problem with ubuntu but Suse 
(at least) is famous for some old bug like the one with blas...

Regards,

N.


Eric Firing wrote:
> Eric,
>
> I don't know much about setup and installation, but based on your 
> message and a quick look at the relevant part of setupext.py, my guess 
> is that there is something odd about your tk/tcl installation.  The 
> code is trying to use methods provided by Tkinter to automatically 
> locate the header files.  This code must work for nearly everyone, or 
> else there would be many problem reports like yours.
>
> You might be able to track down the problem by executing the header 
> discovery code in find_tcltk line-by-line from the interpreter, or by 
> adding lots of print statements.
>
> Did you by any chance manually install your tk/tcl header files, as 
> opposed to installing them as part of a *-devel rpm?  I am assuming 
> you are running Linux.
>
> Eric
>
> Eric Emsellem wrote:
>> Hi,
>>
>> please let me know if this remark is irrelevant:
>>  this is the third email mentioning that I still have to add one line 
>> in setupext.py in order for matplotlib to be installed without 
>> crashing (see below).
>> For some reason this 'bug' (if it is one?) is present since many 
>> versions and has not been solved for me although I have tried to send 
>> emails regularly regarding this item. Please let me know if this is 
>> normal or if it is just something not set up right for me.
>>
>> ALSO: I posted a request for help on mplot3d twice, but got no 
>> feedback. Any help is welcome!
>>
>> thanks in advance,
>> Eric
>>
>> P.S.: here is what I have to do everytime I wish to install a new 
>> version of matplotlib:
>>
>> # add one line in setupext.py:
>>
>>         o.tcl_inc="/usr/include/"
>>
>> # before
>>         if not os.path.exists(o.tcl_inc):
>>             print 'cannot find tcl/tk headers. giving up.'
>>             sys.exit()
>>     return o
>>
>>
>> -------------------------------------------------------
>> Using Tomcat but need to do more? Need to support web services, 
>> security?
>> Get stuff done quickly with pre-integrated technology to make your 
>> job easier
>> Download IBM WebSphere Application Server v.1.0.1 based on Apache 
>> Geronimo
>> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job 
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache 
> Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

After "from pylab import *", log(0.0) gives

Warning: Encountered divide by zero(s)  in log

Where is this behaviour documented?   What does it mean?   Why should =
log(0.0) divide by zero?   What does the plural "zeros" refer to?   Why =
break the IEEE standard that log(0.0) should silently return -Inf?

Keith

PS: at http://matplotlib.sourceforge.net/pylab_commands.html, "it's =
size" should be "its size."

I believe the installation problem that Eric points must be a little more 
widespread than acknowledged here. I for one have to add the line he mentions to 
my setupext.py file each time I install matplotlib. The first time I had this 
problem I found the solution after googling around, so others have had this 
problem before.

More importantly, I am running matplotlib under SUSE 10.0, and all my libraries 
have been installed from the RPMs provided by Suse; ie, having tcl headers in 
"/usr/include/" mustn't be that unusual.

Antonio



John Hunter wrote:
>>>>>> "Eric" == Eric Emsellem <ems...@ob...> writes:
> 
>     Eric> Hi, please let me know if this remark is irrelevant: this is
>     Eric> the third email mentioning that I still have to add one line
>     Eric> in setupext.py in order for matplotlib to be installed
>     Eric> without crashing (see below).  For some reason this 'bug'
>     Eric> (if it is one?) is present since many versions and has not
>     Eric> been solved for me although I have tried to send emails
>     Eric> regularly regarding this item. Please let me know if this is
>     Eric> normal or if it is just something not set up right for me.
> 
>     Eric> ALSO: I posted a request for help on mplot3d twice, but got
>     Eric> no feedback. Any help is welcome!
> 
> 
> mplot3 is not in any official matplotlib version -- it resides in svn 
> and is awaiting integration.  I took a stab at this after your last
> email but ran into difficulties.  Thus you should consider mplot3d
> unsupported until we make a release that includes it.
> 
>     Eric> thanks in advance, Eric
> 
>     Eric> P.S.: here is what I have to do everytime I wish to install
>     Eric> a new version of matplotlib:
> 
>     Eric> # add one line in setupext.py:
> 
>     Eric>          o.tcl_inc="/usr/include/"
> 
> 
> The problem is that this fixes the bug for your system, but is not a
> general solution.  Other platforms will put their tcl/tk libs in a
> different place.  So we need a generic solution which will work across
> platforms, which is why I haven't committed this.  Granted
> /usr/include is a standard location, and this would be a good fallback
> to try if the current method fails (it fails for you and works for
> most -- perhaps some tcl config app is missing on your system?).
> 
> JDH
> 
> 
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On Mon, 24 Apr 2006, Christopher Barker apparently wrote: 
> I think it might use dvips or something to do that. rather than 
> reading and rendering the DVI itself. 

That is not my understanding,
which however is limited.

Cheers,
Alan Isaac

PS I have tried to reopen the discussion with the PyX 
developers.  I'll post any useful outcomes.

John Hunter wrote:
>>>>>> "Alan" == Alan G Isaac <ai...@am...> writes:
>     Alan> Is this the PyX approach?  I ask that in hopes there may
>     Alan> synergies that save some developer time.
> 
> pyx is GPL -- we've contacted them about relicensing their code but
> got nowhere.

I didn't think it was the PyX approach anyway, but I just took another 
look. Does anyone know for sure? In PyX's case, it is going from TeX to 
PS, so I think it might use dvips or something to do that. rather than 
reading and rendering the DVI itself.

Speaking of which, perhaps pdfTeX is an option. I think pdfTeX generates 
pdf directly, which we could either just use, or maybe parse. I don't 
know if that would be harder or easier than DVI though.

-Chris




-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

>>>>> "Alan" == Alan G Isaac <ai...@am...> writes:
    Alan> Is this the PyX approach?  I ask that in hopes there may
    Alan> synergies that save some developer time.

pyx is GPL -- we've contacted them about relicensing their code but
got nowhere.

JDH

Christopher Barker wrote:

> PyShell is good option, if you can get ipython to work with it.


oops, I meant PyCrust. Sorry.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

> Darren Dale wrote: 
>> The basic approach is to extract the font layout information from the 
>> dvi files. LaTeX could be the only dependency. 


On Mon, 24 Apr 2006, Christopher Barker apparently wrote: 
> Wonderful idea! I've been hoping someone would write this for ages. This 
> could lead to using TeX as a layout engine for creating Reports, etc. in 
> all kinds of other apps as well. 


Is this the PyX approach?
I ask that in hopes there may
synergies that save some developer time.

Cheers,
Alan Isaac

Darren Dale wrote:
> The basic approach is to extract the font layout information from the
> dvi files. LaTeX could be the only dependency.

Wonderful idea! I've been hoping someone would write this for ages. This 
could lead to using TeX as a layout engine for creating Reports, etc. in 
all kinds of other apps as well.

I'm looking forward to seeing how it works out.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

Matt Newville wrote:

> I think it would be easy to write a wxPython App (even with
> PythonCard) that provided
> a command-line which was passed off to IPython (say, on an Enter
> event).

PyShell is good option, if you can get ipython to work with it.

-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...