matplotlib-users Mailing List for matplotlib

John Hunter wrote:
>>>>>> "Eric" == Eric Firing <ef...@ha...> writes:
> Eric> We did decide to make linewidth=0 suppress drawing the line,
> Eric> and this is in the ps backend. Most likely it is not in all
> Eric> backends, but it is in pdf, svg, and agg as well as ps. In
> 
> Would it not be better to catch this at the Artist level and either
> not make a call to the backend, or force the edgecolor=facecolor
> before calling the backend? Then the backend can remain blissfully
> ignorant...
No, this strategy won't work for two reasons:
1) edgecolor=facecolor is not the same as not drawing the edge; it leads 
to artifacts that can be very damaging.
2) the backends implement patch rendering which inherently includes both 
the edge and the fill, so we can't avoid having mechanisms for telling 
the backend whether to stroke the former and/or paint the latter.
Eric
> 
> JDH

>>>>> "Eric" == Eric Firing <ef...@ha...> writes:
 Eric> We did decide to make linewidth=0 suppress drawing the line,
 Eric> and this is in the ps backend. Most likely it is not in all
 Eric> backends, but it is in pdf, svg, and agg as well as ps. In
Would it not be better to catch this at the Artist level and either
not make a call to the backend, or force the edgecolor=facecolor
before calling the backend? Then the backend can remain blissfully
ignorant...
JDH

John Hunter wrote:
> Whether or not linewidth=0 is respected may be backend dependent. In
> postscript it means use the smallest possible line. We have discussed
> overriding this at the mpl level to impose consistency, but am not
> sure what the status of this is. The safest best is to set the
> edgecolor and facecolor to be identical.
John,
We did decide to make linewidth=0 suppress drawing the line, and this is 
in the ps backend. Most likely it is not in all backends, but it is in 
pdf, svg, and agg as well as ps. In the ps backend it is in the 
_draw_ps() method, so unless there is a code path that does not use this 
method, it should work.
Sergio, I deleted your original message so I don't remember how much 
info you gave. If linewidth=0 is not suppressing edge drawing and you 
are using mpl 0.87.7 or svn, then please provide a simple script 
illustrating the problem and say what backend and mpl version you are using.
Eric

>>>>> "Sergio" == Sergio Drumond Ventura <da...@gm...> writes:
 Sergio> Hi, I am a new user of matplotlib. First of all I would
 Sergio> like to say that the people working on matplotlib is doing
 Sergio> a great job.
 Sergio> Well, I have tried to use the 'fill', but I cannot get rid
 Sergio> of the lines. I have tried to use 'linewidth=0' at the
 Sergio> end, but it did not work.
 Sergio> Could somebody, please, tell me if I am doing something
 Sergio> wrong ?
Whether or not linewidth=0 is respected may be backend dependent. In
postscript it means use the smallest possible line. We have discussed
overriding this at the mpl level to impose consistency, but am not
sure what the status of this is. The safest best is to set the
edgecolor and facecolor to be identical.
JDH

>>>>> "Steve" == Steve Kacenjar <kac...@co...> writes:
 Steve> Just installed FC6 onto AMD 64 bit processor and have
 Steve> installed Python. I have also installed numpy and
 Steve> matplotlib however when I run IPython and begin importing
 Steve> items I run into a problem when try to import pylab.
 Steve> Specifically I receive the following dump:
 Steve> ImportError: No module named _ns_backend_gdk
Or a 2nd thought -- it may be that pygtk or the gtk-devel environment
was not in place when you built. Even if they are installed,
matplotlib requires a connection to the X11 server when building the
gtk backend, something that is not always available when building as
root.
So 
 1) make sure you have the proper numpy installed, as in my previous
 post
 2) make sure you can do
 >>> import gtk
 in the same environment in which you are building and installing
 mpl
 3) make sure you have the gtk-devel headers and libs installed, as
 well as the other devel libs listed on the installing pages of
 the web site.
 4) capture stdout and stderr when you build, and post it if you
 encounter further problems
 > python setup.py build >& build.out
JDH

>>>>> "Steve" == Steve Kacenjar <kac...@co...> writes:
 Steve> Just installed FC6 onto AMD 64 bit processor and have
 Steve> installed Python. I have also installed numpy and
 Steve> matplotlib however when I run IPython and begin importing
 Steve> items I run into a problem when try to import pylab.
 Steve> Specifically I receive the following dump:
 Steve> ImportError: No module named _ns_backend_gdk
 Steve> I am new to this and would appreciate any help in resolving
 Steve> this problem. Thanks.
this looks like numpy was not available at compile time. Make sure
you have a compatible numpy and mpl (eg the latest release of both or
the latest svn of both) and then do a clean build (eg 'rm -rf build').
JDH

>>>>> "Eric" == Eric Firing <ef...@ha...> writes:
 Eric> I don't think so--at least, I have not figured it out after
 Eric> a few minutes of looking around and trying to understand how
 Eric> things work. I think polar plotting needs quite a bit of
 Eric> work. In particular, I suspect the lack of a "set_rmax"
 Eric> method is not an oversight but a consequence of limitations
 Eric> in the present implementation. I will look at it some more
 Eric> and see if I can improve it. This might be long-term, not
 Eric> quick.
In the next month or so I plan to spend some time fixing up the
limitations of polar, so I'll flag this and take a look at it then.
JDH

>>>>> "Edin" == Edin Salkovic <edi...@gm...> writes:
 Edin> Hi, I'm having problems with building mpl from SVN on WinXP,
 Edin> Python 2.5, latest numpy. I'm using MinGW from the
 Edin> enthought edition. I remember building it without problems
 Edin> just few weeks ago. The build process finishes without
 Edin> error. Then, when I try:
 Edin> from pylab import *
 Edin> it crashes
 Edin> import matplotlib
Did you rm -rf your build dir (and sometimes site-packages/matplotlib)
before rebuilding. This is usually the cause of such crashes.
JDH