[Matplotlib-users] Problems with matplotlib

[<na...@te...>]

Greetings.

I'm new to matplotlib and to this list. I know it is a
good thing to wait a little before posting, but I need
some work done quickly, so I had to post a message as
soon as I could. I also couldn't find in the list
archives a solution to my problems, so I had to ask.
Bear with me, please.

I must say that I'm very impressed with the overall
quality of the figures generated with matplotlib, and
the possibility to add TeX expression as labels is a
fantastic feature that I have never even thought about
(but now I can't live without).

I had some problems though, which I report below. Some
of them I could solve (by hacking a little), others I
didn't.

* I use TK backend, but also tried wx. They didn't work
  well: in the TK backend, in interactive mode, the fi-
  gure take a lot of time to be rendered, and sometimes
  it does not. The same with wx. I looked into the
  code of 'show' method in backend_tkagg.py, and did
  some changes. The Tk.mainloop() was commented, I
  uncommented it. I also changed the interactive mode
  to False. By doing this, I lost the interactive
  environment, but at least the figure shows quickly
  when I use the 'show' command. I don't know if this
  has been an issue for everyone, or if it is a problem
  with compatibility with what I have installed here
  (it is Python 2.3.2, Windows ME, Tk 8.4, if I'm right),
  but if there is a better solution to it, I'd like to
  hear.

* Also in the backend_tkagg.py there was two lines with
  wrong identation. Line 181 (method show of
  FigureManagerTkAgg class and line 416 (method update
  of NavigationToolbar class) were not idented correctly.
  I tried identing them and everything seems to be working
  fine. I might have changed something that I shouldn't,
  though.

* Well, here is my real problem. I need to generate EPS
  files of my pictures for use with TeX (MiKTeX). I'm
  having a lot of trouble with it (this includes problems
  with the GS conversion tool, but that is beyond the
  scope of this list). Basically, using TkAgg or the
  savefig method, I can generate a PS file with the
  figure. But the problem is that the figure does not
  show well - it is bigger than the generated page, so a
  very significant part of the figure is cut out. Is
  there anyway to resize the picture or tell matplotlib
  to fit the picture within specified dimensions (like
  some and some centimeters or inches height and width)?
  Resizing the drawing window _did_ solve the problem,
  but then I can't generate the figures by scripting -
  I would have to manually resize everyone of them.

* The EPS problem I might solve with a little hack, and
  generate an EPS file directly from my scripts. If this
  turns out to work well, I will contribute the code, of
  course, if the developers would want it.

* Another very important thing to me is to generate the
  plot of discrete functions (like the Matlab 'mesh'
  command). Is there anyway to do that with matplotlib?
  If not, can I suggest it as a feature request? :)

That's all for now. Thanks for the great software.

---
José Alexandre Nalon
na...@te...