Gtk.Layout instead of Gtk.DrawingArea

The patch works OK when I tried it, and the example displayed a button
in the FigureCanvas. How is it extremely useful to place a widget on top
of a FigureCanvas?

I think a figure widget should just show figures, extra widgets can
always be placed outside the figure widget.
Gtk.DrawingArea is a simple, fast widget. Gtk.Layout (as used by
GnomeCanvas) is more complicated and has a higher overhead.

This week I read that the widget item in gnome-canvas does not work very
well
http://blogs.gnome.org/view/alexl/2005/09/14/0

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

···

On Wed, 2005-09-14 at 20:32 -0700, John Hunter wrote:

        OK, great. Steve could you take a look at this when you get a
        minute
        and give some feedback about whether you think this is a god
        idea.

Currently matplotlib already draws its widgets on the canvas, so this would allow a method to instead use native widgets in place of the matplotlib widgets when possible. Using native widgets has the advantage of response time and having a larger library at one's disposal. Even without the mapping of matplotlib->native widgets, it can be useful to easily put more controls to allow the plot to be altered (e.g. playing a movie, rerunning a sim, etc.) This can also be accomplished (at least with GTK) by adding another toolbar, but isn't always optimal.

Is there an easy method to place widgets outside of the figure widget? Also, does this mean that matplotlib widgets should also not be drawn on the canvas?

I checked on the link you gave, and it appears the problem occurs when scrolling is involved. Currently with how things are done, I don't think this will happen.

Abe

Steve Chaplin wrote:

···

On Wed, 2005-09-14 at 20:32 -0700, John Hunter wrote:

       OK, great. Steve could you take a look at this when you get a
       minute
       and give some feedback about whether you think this is a god
       idea.
   
The patch works OK when I tried it, and the example displayed a button
in the FigureCanvas. How is it extremely useful to place a widget on top
of a FigureCanvas?

I think a figure widget should just show figures, extra widgets can
always be placed outside the figure widget.
Gtk.DrawingArea is a simple, fast widget. Gtk.Layout (as used by
GnomeCanvas) is more complicated and has a higher overhead.

This week I read that the widget item in gnome-canvas does not work very
well
http://blogs.gnome.org/view/alexl/2005/09/14/0

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
Matplotlib-devel mailing list
Matplotlib-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

Currently matplotlib already draws its widgets on the canvas, so this
would allow a method to instead use native widgets in place of the
matplotlib widgets when possible. Using native widgets has the advantage
of response time and having a larger library at one's disposal. Even
without the mapping of matplotlib->native widgets, it can be useful to
easily put more controls to allow the plot to be altered (e.g. playing a
movie, rerunning a sim, etc.) This can also be accomplished (at least
with GTK) by adding another toolbar, but isn't always optimal.

Is there an easy method to place widgets outside of the figure widget?

Use the usual GTK+ widget placement methods - create a gtk.Window, add a
VBox/HBox/Table etc and place the figure widget and other widgets in the
container. This would be using the matplotlib OO interface like in
examples/embedding_in_gtk2.py

Also, does this mean that matplotlib widgets should also not be drawn on
the canvas?

Should matplotlib have turned itself into a widget library / GUI toolkit?
I know that for a long time John resisted the temptation to add widgets
to matplotlib and wanted matplotlib to focus on being a plotting
library. I agree with this view and think the danger now is that
matplotlib will become too big and bloated and harder to install (like
the old SciPy?).

I would prefer to see a 'matplotlib-core' which is a minimal module that
focuses purely on plotting graphs. Other optional modules (like
'matplotlib-toolkits', 'matplotlib-widgets', etc) could then extend
'matplotlib-core' by providing extra features.

I checked on the link you gave, and it appears the problem occurs when
scrolling is involved. Currently with how things are done, I don't think
this will happen.

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

···

On Sun, 2005-09-18 at 16:45 -0400, Abraham Schneider wrote:

Steve,

Darren Dale raised a question offline that I think will be of general interest to mpl devels, and that you are the person to answer:

How do you see the future of a cairo backend as a prospective replacement for most or all of the primary mpl backends?

What would need to be completed in cairo? Based on the cairo web page, I get the impression that quite a bit is still missing, including eps generation and genuine vector ps. But maybe the web page is just out of date.

What would need to be done in mpl, and how hard would it be?

Would mpl get slower if everything went through cairo?

Any other considerations?

Thanks.

Eric

Steve,

One more question: how does the image quality of cairo compare to Agg? Is the antialiasing as good?

Thanks.

Eric

Steve,

Darren Dale raised a question offline that I think will be of general
interest to mpl devels, and that you are the person to answer:

How do you see the future of a cairo backend as a prospective
replacement for most or all of the primary mpl backends?

I think cairo could potentially be used to replace the pdf, ps, svg
and gdk/gtk backends which would unify most of the backends and simplify
a lot of the code.

What would need to be completed in cairo? Based on the cairo web page, I
get the impression that quite a bit is still missing, including eps
generation and genuine vector ps. But maybe the web page is just out of
date.

Which web page is out of date? Where does it mention eps and ps, I
couldn't find it.

My general impression of the cairo "surfaces" is:
ImageSurface/png - support is very good
gtk/xlib - support is very good
ps/pdf/svg are usable but less mature and still developing so there may
            be occasional problems drawing specific items
ps - it used to embed bitmap images but now most output is vector based
eps - is not supported yet, but may be in a future version

What would need to be done in mpl, and how hard would it be?

The cairo backend can already be used for png, ps, pdf and gtk output so
I don't think there would be much to do. Mostly, it needs testing -
running all the mpl examples and checking the output looks OK.

Would mpl get slower if everything went through cairo?

Not sure, you would need to run cairo and test it. It used to be much slower
than Agg but more recent versions have had many optimisations applied and
the difference is much smaller now.

Any other considerations?

Some parts of mpl are Agg-specific and other parts (the whole drawing model)
are designed around the gdk drawing style - this makes things harder and
inefficient when using cairo.

One more question: how does the image quality of cairo compare to
Agg?
Is the antialiasing as good?

Image quality looks OK to me, but I'm no expert.

The web browser Firefox 3.0 (due to be released early in 2007) will use
cairo for all rendering. Firefox requires a high level of graphics
performance and the upcoming cairo 1.4 release is expected to provide
that.

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

···

On Sat, 2007-01-06 at 09:23 -1000, Eric Firing wrote:
On Sat, 2007-01-06 at 09:36 -1000, Eric Firing wrote:

Hi Steve, Eric, John,

> Steve,
>
> Darren Dale raised a question offline that I think will be of general
> interest to mpl devels, and that you are the person to answer:
>
> How do you see the future of a cairo backend as a prospective
> replacement for most or all of the primary mpl backends?

I think cairo could potentially be used to replace the pdf, ps, svg
and gdk/gtk backends which would unify most of the backends and simplify
a lot of the code.

This would really be a plus. The option to use latex for text layout would
have to be completely reworked, if it could be supported at all. Not that
this is a critical feature, but in my opinion it is still necessary at this
point for producing the highest quality plots for publication.

> What would need to be completed in cairo? Based on the cairo web page, I
> get the impression that quite a bit is still missing, including eps
> generation and genuine vector ps. But maybe the web page is just out of
> date.

Which web page is out of date? Where does it mention eps and ps, I
couldn't find it.

My general impression of the cairo "surfaces" is:
ImageSurface/png - support is very good
gtk/xlib - support is very good
ps/pdf/svg are usable but less mature and still developing so there may
            be occasional problems drawing specific items
ps - it used to embed bitmap images but now most output is vector based
eps - is not supported yet, but may be in a future version

> What would need to be done in mpl, and how hard would it be?

The cairo backend can already be used for png, ps, pdf and gtk output so
I don't think there would be much to do. Mostly, it needs testing -
running all the mpl examples and checking the output looks OK.

I had to alter the following lines from backend_gtkcairo, from

import matplotlib.backends.backend_cairo as be_cairo
from matplotlib.backends.backend_gtk import *

to

import backend_cairo as be_cairo
from backend_gtk import *

in order to prevent the following traceback:

Traceback (most recent call last):
  File "/usr/bin/ipython", line 27, in ?
    IPython.Shell.start().mainloop()
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 1034, in
start
    return shell(user_ns = user_ns)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 945, in
__init__
    shell_class=MatplotlibMTShell)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 622, in
__init__
    on_kill=[mainquit])
  File "/usr/lib64/python2.4/site-packages/IPython/ipmaker.py", line 90, in
make_IPython
    embedded=embedded,**kw)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 506, in
__init__
    user_ns,b2 = self._matplotlib_config(name,user_ns)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 397, in
_matplotlib_config
    from matplotlib import backends
  File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
line 55, in ?
    new_figure_manager, draw_if_interactive, show = pylab_setup()
  File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
line 23, in pylab_setup
    globals(),locals(),[backend_name])
  
File "/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py",
line 11, in ?
    import matplotlib.backends.backend_cairo as be_cairo
AttributeError: 'module' object has no attribute 'backends'

I only had a short time to work with backend_gtkcairo, but a couple of things
caught my attention. mathtext support seemed to need some improvement, it
looked like it was rendered as an image rather than as text. Also,
imshow(rand(100,100)) looked very different with gtkcairo and gtkagg, (maybe
because the rgba ordering is different in agg and cairo? I'm not sure this is
even the case, I'm taking a stab in the dark.)

> Would mpl get slower if everything went through cairo?

Not sure, you would need to run cairo and test it. It used to be much
slower than Agg but more recent versions have had many optimisations
applied and the difference is much smaller now.

> Any other considerations?

Some parts of mpl are Agg-specific and other parts (the whole drawing
model) are designed around the gdk drawing style - this makes things harder
and inefficient when using cairo.

I'm just thinking out loud here, brainstorming, but may be getting completely
ahead of myself. Feel free to tell me so. If we made a mpl-pre1.0 branch, and
reconstructed the drawing model, how much work are we talking about? Since
gtk includes cairo now, couldn't a single gtk backend replace both gtk and
gtkagg, while retaining the speed of the pure gtk backend?

···

On Sunday 07 January 2007 04:44, Steve Chaplin wrote:

On Sat, 2007-01-06 at 09:23 -1000, Eric Firing wrote:

On Sat, 2007-01-06 at 09:36 -1000, Eric Firing wrote:
> One more question: how does the image quality of cairo compare to
> Agg?
> Is the antialiasing as good?

Image quality looks OK to me, but I'm no expert.

The web browser Firefox 3.0 (due to be released early in 2007) will use
cairo for all rendering. Firefox requires a high level of graphics
performance and the upcoming cairo 1.4 release is expected to provide
that.

Darren Dale wrote:

Hi Steve, Eric, John,

How do you see the future of a cairo backend as a prospective
replacement for most or all of the primary mpl backends?

I think cairo could potentially be used to replace the pdf, ps, svg
and gdk/gtk backends which would unify most of the backends and simplify
a lot of the code.

This would really be a plus. The option to use latex for text layout would have to be completely reworked, if it could be supported at all. Not that this is a critical feature, but in my opinion it is still necessary at this point for producing the highest quality plots for publication.

Playing around with using Cairo for something else, I once wrote some
code to convert dvi to Cairo. The code I wrote probably isn't suitable
for use by matplotlib, but it wasn't particularly difficult to
write. The most difficult thing was finding the correct fonts - but I know people on this list have a lot of experience with that kind of problem anyway. It is possible the code would need to be in c/c++ for speed - my code was anyway so I don't know.

I would offer to write something myself, but I'm leaving the field in
which I use matplotlib around now and I wouldn't be able to maintain
anything I wrote.

Nicholas

···

On Sunday 07 January 2007 04:44, Steve Chaplin wrote:

On Sat, 2007-01-06 at 09:23 -1000, Eric Firing wrote:

Would you be willing to contribute this code to mpl? If so, please forward it
to me off list. I would really like to have a look.

Darren

···

On Monday 08 January 2007 11:51, Nicholas Young wrote:

Darren Dale wrote:
> Hi Steve, Eric, John,
>
> On Sunday 07 January 2007 04:44, Steve Chaplin wrote:
>> On Sat, 2007-01-06 at 09:23 -1000, Eric Firing wrote:
>>> How do you see the future of a cairo backend as a prospective
>>> replacement for most or all of the primary mpl backends?
>>
>> I think cairo could potentially be used to replace the pdf, ps, svg
>> and gdk/gtk backends which would unify most of the backends and simplify
>> a lot of the code.
>
> This would really be a plus. The option to use latex for text layout
> would have to be completely reworked, if it could be supported at all.
> Not that this is a critical feature, but in my opinion it is still
> necessary at this point for producing the highest quality plots for
> publication.

Playing around with using Cairo for something else, I once wrote some
code to convert dvi to Cairo. The code I wrote probably isn't suitable
for use by matplotlib, but it wasn't particularly difficult to
write. The most difficult thing was finding the correct fonts - but I
know people on this list have a lot of experience with that kind of
problem anyway. It is possible the code would need to be in c/c++ for
speed - my code was anyway so I don't know.

I would offer to write something myself, but I'm leaving the field in
which I use matplotlib around now and I wouldn't be able to maintain
anything I wrote.

The original matplotlib code is correct, you should be editing IPython
and correcting their bug!
Matplotlib does use a lot of relative imports which I think is bad
style.

See PEP 8 "Style Guide for Python Code"
http://www.python.org/dev/peps/pep-0008/

    - Relative imports for intra-package imports are highly discouraged.
      Always use the absolute package path for all imports.
      Even now that PEP 328 [7] is fully implemented in Python 2.5,
      its style of explicit relative imports is actively discouraged;
      absolute imports are more portable and usually more readable.

There was a recent "Coding Guide" thread on this list (which I admit I just
skimmed through). Instead of reinventing the wheel, how about stating at the
top of CODING_GUIDE that PEP 8 is the default style for matplotlib, and the
following notes give in-depth matplotlib examples (or possibly override
PEP 8 if necessary).

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

···

On Mon, 2007-01-08 at 11:24 -0500, Darren Dale wrote:

> > What would need to be done in mpl, and how hard would it be?
>
> The cairo backend can already be used for png, ps, pdf and gtk output so
> I don't think there would be much to do. Mostly, it needs testing -
> running all the mpl examples and checking the output looks OK.

I had to alter the following lines from backend_gtkcairo, from

import matplotlib.backends.backend_cairo as be_cairo
from matplotlib.backends.backend_gtk import *

to

import backend_cairo as be_cairo
from backend_gtk import *

in order to prevent the following traceback:

Traceback (most recent call last):
  File "/usr/bin/ipython", line 27, in ?
    IPython.Shell.start().mainloop()
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 1034, in
start
    return shell(user_ns = user_ns)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 945, in
__init__
    shell_class=MatplotlibMTShell)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 622, in
__init__
    on_kill=[mainquit])
  File "/usr/lib64/python2.4/site-packages/IPython/ipmaker.py", line 90, in
make_IPython
    embedded=embedded,**kw)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 506, in
__init__
    user_ns,b2 = self._matplotlib_config(name,user_ns)
  File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 397, in
_matplotlib_config
    from matplotlib import backends
  File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
line 55, in ?
    new_figure_manager, draw_if_interactive, show = pylab_setup()
  File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
line 23, in pylab_setup
    globals(),locals(),[backend_name])
  
File "/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py",
line 11, in ?
    import matplotlib.backends.backend_cairo as be_cairo
AttributeError: 'module' object has no attribute 'backends'

Well, I'd be delighted to correct the ipython bug, if only I
understood exactly what the problem was... As far as I can see, that
code in ipython is simply calling

        # Initialize matplotlib to interactive mode always
        import matplotlib
        from matplotlib import backends

inside a function (the _matplotlib_config method). I don't see a bug
in that, but if you provide some pointers, I'll be happy to fix any
issues that are on ipython's side of the fence.

Cheers,

f

···

On 1/10/07, Steve Chaplin <stevech1097@...49...> wrote:

On Mon, 2007-01-08 at 11:24 -0500, Darren Dale wrote:

> I had to alter the following lines from backend_gtkcairo, from
>
> import matplotlib.backends.backend_cairo as be_cairo
> from matplotlib.backends.backend_gtk import *
>
> to
>
> import backend_cairo as be_cairo
> from backend_gtk import *
>
> in order to prevent the following traceback:
>
> Traceback (most recent call last):
> File "/usr/bin/ipython", line 27, in ?
> IPython.Shell.start().mainloop()
> File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 1034, in
> start
> return shell(user_ns = user_ns)
> File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 945, in
> __init__
> shell_class=MatplotlibMTShell)
> File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 622, in
> __init__
> on_kill=[mainquit])
> File "/usr/lib64/python2.4/site-packages/IPython/ipmaker.py", line 90, in
> make_IPython
> embedded=embedded,**kw)
> File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 506, in
> __init__
> user_ns,b2 = self._matplotlib_config(name,user_ns)
> File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 397, in
> _matplotlib_config
> from matplotlib import backends
> File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
> line 55, in ?
> new_figure_manager, draw_if_interactive, show = pylab_setup()
> File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
> line 23, in pylab_setup
> globals(),locals(),[backend_name])
>
> File "/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py",
> line 11, in ?
> import matplotlib.backends.backend_cairo as be_cairo
> AttributeError: 'module' object has no attribute 'backends'

The original matplotlib code is correct, you should be editing IPython
and correcting their bug!

cairo mathtext uses a method copied from gdk/gtk and does render an
image. It needs updating to render text.

imshow does look different on cairo and agg, and yes, It looks like an
image format problem. cairo uses ARGB32 with pre-multiplied alpha, and
the ARGB order depends on whether the machine is little- of big-endian.

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

···

On Mon, 2007-01-08 at 11:24 -0500, Darren Dale wrote:

I only had a short time to work with backend_gtkcairo, but a couple of things
caught my attention. mathtext support seemed to need some improvement, it
looked like it was rendered as an image rather than as text. Also,
imshow(rand(100,100)) looked very different with gtkcairo and gtkagg, (maybe
because the rgba ordering is different in agg and cairo? I'm not sure this is
even the case, I'm taking a stab in the dark.)

> > "/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcair
> >o.py", line 11, in ?
> > import matplotlib.backends.backend_cairo as be_cairo
> > AttributeError: 'module' object has no attribute 'backends'
>
> The original matplotlib code is correct, you should be editing IPython
> and correcting their bug!

Well, I'd be delighted to correct the ipython bug, if only I
understood exactly what the problem was... As far as I can see, that
code in ipython is simply calling

        # Initialize matplotlib to interactive mode always
        import matplotlib
        from matplotlib import backends

inside a function (the _matplotlib_config method). I don't see a bug
in that, but if you provide some pointers, I'll be happy to fix any
issues that are on ipython's side of the fence.

I've been looking at this, but haven't made much progress. Try this with
backend:gtkcairo in matplotlibrc:

$ python

__import__('matplotlib.backends.backend_ps', globals(),\

locals(),['backend_ps'])

output:
<module 'matplotlib.backends.backend_ps'
from '/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_ps.pyc'>

$ ipython

In [1]: __import__('matplotlib.backends.backend_ps', globals(),\
locals(),['backend_ps'])

output:

···

On Thursday 11 January 2007 02:01, Fernando Perez wrote:

On 1/10/07, Steve Chaplin <stevech1097@...49...> wrote:
> On Mon, 2007-01-08 at 11:24 -0500, Darren Dale wrote:

---------------------------------------------------------------------------
exceptions.AttributeError Traceback (most recent
call last)

/home/darren/<ipython console>

/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py
     54
     55 # a hack to keep old versions of ipython working with mpl
     56 if 'IPython.Shell' in sys.modules:
---> 57 new_figure_manager, draw_if_interactive, show = pylab_setup()
     58

/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py in
pylab_setup()
     24 time.sleep(1)
     25 backend_mod = __import__('matplotlib.backends.'+backend_name,
---> 26 globals(),locals(),[backend_name])
     27
     28 # Things we pull in from all backends

/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py
      7 import cairo.gtk
      8
----> 9 import matplotlib.backends.backend_cairo as be_cairo
     10 from matplotlib.backends.backend_gtk import *
     11

AttributeError: 'module' object has no attribute 'backends'

If you change backend:gtkagg in matplotlibrc, then ipython doesnt complain.
Very strange.

I found a workaround, and committed it. I don't understand what the root of
the problem was, probably because the AttributeError at the end of the
Traceback is misleading. Changing this:

import matplotlib.backends.backend_cario as be_cairo

to this:

from matplotlib.backends import backend_cairo

and updating references to be_cairo was all that was needed. The fix is in svn
2979.

Darren

···

On Thursday 11 January 2007 16:12, Darren Dale wrote:

$ ipython

In [1]: __import__('matplotlib.backends.backend_ps', globals(),\
locals(),['backend_ps'])

output:
---------------------------------------------------------------------------
exceptions.AttributeError Traceback (most recent
call last)

/home/darren/<ipython console>

/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py
     54
     55 # a hack to keep old versions of ipython working with mpl
     56 if 'IPython.Shell' in sys.modules:
---> 57 new_figure_manager, draw_if_interactive, show = pylab_setup()
     58

/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py in
pylab_setup()
     24 time.sleep(1)
     25 backend_mod = __import__('matplotlib.backends.'+backend_name,
---> 26 globals(),locals(),[backend_name])
     27
     28 # Things we pull in from all backends

/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py
      7 import cairo.gtk
      8
----> 9 import matplotlib.backends.backend_cairo as be_cairo
     10 from matplotlib.backends.backend_gtk import *
     11

AttributeError: 'module' object has no attribute 'backends'

Darren reported a "bug" in backend_gtkcairo.py which he has "fixed". My
view is that the lines
from matplotlib.backends import backend_cairo
from matplotlib.backends.backend_gtk import *
work fine when called from the Python prompt. They are using the
absolute package path and are correct and should not be "fixed" to use
relative imports (which is bad style). So if there is a bug it is
elsewhere and from the traceback it looked like ipython is involved.

The traceback shows that this code is being executed
# a hack to keep old versions of ipython working with mpl
if 'IPython.Shell' in sys.modules:
    new_figure_manager, draw_if_interactive, show = pylab_setup()

which is ipython-specific code.
I don't usually use IPython, but I installed it today and ran a few
simple matplotlib plots with the GTKCairo backend and they worked OK! So
I can't offer more info, perhaps Darren can produce a minimal test case
to isolate the problem.

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

···

On Thu, 2007-01-11 at 00:01 -0700, Fernando Perez wrote:

On 1/10/07, Steve Chaplin <stevech1097@...49...> wrote:
> On Mon, 2007-01-08 at 11:24 -0500, Darren Dale wrote:

> > I had to alter the following lines from backend_gtkcairo, from
> >
> > import matplotlib.backends.backend_cairo as be_cairo
> > from matplotlib.backends.backend_gtk import *
> >
> > to
> >
> > import backend_cairo as be_cairo
> > from backend_gtk import *
> >
> > in order to prevent the following traceback:
> >
> > Traceback (most recent call last):
> > File "/usr/bin/ipython", line 27, in ?
> > IPython.Shell.start().mainloop()
> > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 1034, in
> > start
> > return shell(user_ns = user_ns)
> > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 945, in
> > __init__
> > shell_class=MatplotlibMTShell)
> > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 622, in
> > __init__
> > on_kill=[mainquit])
> > File "/usr/lib64/python2.4/site-packages/IPython/ipmaker.py", line 90, in
> > make_IPython
> > embedded=embedded,**kw)
> > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 506, in
> > __init__
> > user_ns,b2 = self._matplotlib_config(name,user_ns)
> > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line 397, in
> > _matplotlib_config
> > from matplotlib import backends
> > File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
> > line 55, in ?
> > new_figure_manager, draw_if_interactive, show = pylab_setup()
> > File "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
> > line 23, in pylab_setup
> > globals(),locals(),[backend_name])
> >
> > File "/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py",
> > line 11, in ?
> > import matplotlib.backends.backend_cairo as be_cairo
> > AttributeError: 'module' object has no attribute 'backends'
>
> The original matplotlib code is correct, you should be editing IPython
> and correcting their bug!

Well, I'd be delighted to correct the ipython bug, if only I
understood exactly what the problem was... As far as I can see, that
code in ipython is simply calling

        # Initialize matplotlib to interactive mode always
        import matplotlib
        from matplotlib import backends

inside a function (the _matplotlib_config method). I don't see a bug
in that, but if you provide some pointers, I'll be happy to fix any
issues that are on ipython's side of the fence.

They work ok now, after changing the import statement so that it doesnt rename
backend_cairo to be_cairo in the gtkcairo namespace. I can't think of a more
minimal example than the one I already provided: starting IPython without the
pylab flag and importing matplotlib.backends.backend_gtkcairo. I looked at
this for a couple hours last week, and was not able to determine whether it
was an IPython bug or a matplotlib bug that is exposed by IPython's magic. I
do agree that the absolute path imports are correct and are not the source of
the problem. Since a workaround has been committed, I don't think I should
spend more time on this issue (unless another problem emerges).

Darren

···

On Monday 15 January 2007 12:16 am, Steve Chaplin wrote:

On Thu, 2007-01-11 at 00:01 -0700, Fernando Perez wrote:
> On 1/10/07, Steve Chaplin <stevech1097@...49...> wrote:
> > On Mon, 2007-01-08 at 11:24 -0500, Darren Dale wrote:
> > > I had to alter the following lines from backend_gtkcairo, from
> > >
> > > import matplotlib.backends.backend_cairo as be_cairo
> > > from matplotlib.backends.backend_gtk import *
> > >
> > > to
> > >
> > > import backend_cairo as be_cairo
> > > from backend_gtk import *
> > >
> > > in order to prevent the following traceback:
> > >
> > > Traceback (most recent call last):
> > > File "/usr/bin/ipython", line 27, in ?
> > > IPython.Shell.start().mainloop()
> > > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line
> > > 1034, in start
> > > return shell(user_ns = user_ns)
> > > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line
> > > 945, in __init__
> > > shell_class=MatplotlibMTShell)
> > > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line
> > > 622, in __init__
> > > on_kill=[mainquit])
> > > File "/usr/lib64/python2.4/site-packages/IPython/ipmaker.py", line
> > > 90, in make_IPython
> > > embedded=embedded,**kw)
> > > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line
> > > 506, in __init__
> > > user_ns,b2 = self._matplotlib_config(name,user_ns)
> > > File "/usr/lib64/python2.4/site-packages/IPython/Shell.py", line
> > > 397, in _matplotlib_config
> > > from matplotlib import backends
> > > File
> > > "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
> > > line 55, in ?
> > > new_figure_manager, draw_if_interactive, show = pylab_setup()
> > > File
> > > "/usr/lib64/python2.4/site-packages/matplotlib/backends/__init__.py",
> > > line 23, in pylab_setup
> > > globals(),locals(),[backend_name])
> > >
> > > File
> > > "/usr/lib64/python2.4/site-packages/matplotlib/backends/backend_gtkca
> > >iro.py", line 11, in ?
> > > import matplotlib.backends.backend_cairo as be_cairo
> > > AttributeError: 'module' object has no attribute 'backends'
> >
> > The original matplotlib code is correct, you should be editing IPython
> > and correcting their bug!
>
> Well, I'd be delighted to correct the ipython bug, if only I
> understood exactly what the problem was... As far as I can see, that
> code in ipython is simply calling
>
> # Initialize matplotlib to interactive mode always
> import matplotlib
> from matplotlib import backends
>
> inside a function (the _matplotlib_config method). I don't see a bug
> in that, but if you provide some pointers, I'll be happy to fix any
> issues that are on ipython's side of the fence.

Darren reported a "bug" in backend_gtkcairo.py which he has "fixed". My
view is that the lines
from matplotlib.backends import backend_cairo
from matplotlib.backends.backend_gtk import *
work fine when called from the Python prompt. They are using the
absolute package path and are correct and should not be "fixed" to use
relative imports (which is bad style). So if there is a bug it is
elsewhere and from the traceback it looked like ipython is involved.

The traceback shows that this code is being executed
# a hack to keep old versions of ipython working with mpl
if 'IPython.Shell' in sys.modules:
    new_figure_manager, draw_if_interactive, show = pylab_setup()

which is ipython-specific code.
I don't usually use IPython, but I installed it today and ran a few
simple matplotlib plots with the GTKCairo backend and they worked OK! So
I can't offer more info, perhaps Darren can produce a minimal test case
to isolate the problem.

Those lines also work fine from an ipython prompt:

In [1]: from matplotlib.backends import backend_cairo

In [2]: from matplotlib.backends.backend_gtk import *

The point is that for some bizarre reason, /inside/ the
ipython/matplotlib initialization, Darren was seeing a problem. It
may very well be that the real culprit is ipython, but after looking
at the issue I can't see anything, and Darren also tried to understand
it and failed.

So given this, Darren found a solution by modifying matplotlib.
Unless you can suggest a proper fix, I think we'll have for now to
live with this, even if it's in principle not ideal. If none of us
can figure out the real problem, at least Darren's solution works, so
it's better than leaving the bug in place.

Cheers,

f

···

On 1/14/07, Steve Chaplin <stevech1097@...49...> wrote:

Darren reported a "bug" in backend_gtkcairo.py which he has "fixed". My
view is that the lines
from matplotlib.backends import backend_cairo
from matplotlib.backends.backend_gtk import *
work fine when called from the Python prompt. They are using the
absolute package path and are correct and should not be "fixed" to use
relative imports (which is bad style). So if there is a bug it is
elsewhere and from the traceback it looked like ipython is involved.

The traceback shows that this code is being executed
# a hack to keep old versions of ipython working with mpl
if 'IPython.Shell' in sys.modules:
    new_figure_manager, draw_if_interactive, show = pylab_setup()

which is ipython-specific code.
I don't usually use IPython, but I installed it today and ran a few
simple matplotlib plots with the GTKCairo backend and they worked OK! So
I can't offer more info, perhaps Darren can produce a minimal test case
to isolate the problem.

Those lines also work fine from an ipython prompt:

In [1]: from matplotlib.backends import backend_cairo

In [2]: from matplotlib.backends.backend_gtk import *

They work, but they are not relevant to this problem. Darren has
confused the issue by applying a fix to backend_gtkcairo.py (with no
CHANGELOG comment) while you and I are investigating the problem. So we
are executing the workaround command and are unable to reproduce the
bug.

However, if I put the line
import matplotlib.backends.backend_cairo as be_cairo
back into backend_gtkcairo.py I can now see the problem.

$ python
Python 2.4.3 (#1, Oct 1 2006, 18:00:19)
[GCC 4.1.1 20060928 (Red Hat 4.1.1-28)] on linux2
Type "help", "copyright", "credits" or "license" for more information.

import matplotlib.backends.backend_gtkcairo

$ ipython
Python 2.4.3 (#1, Oct 1 2006, 18:00:19)
Type "copyright", "credits" or "license" for more information.

IPython 0.7.2 -- An enhanced Interactive Python.
? -> Introduction to IPython's features.
%magic -> Information about IPython's 'magic' % functions.
help -> Python's own help system.
object? -> Details about 'object'. ?object also works, ?? prints more.

In [1]: import matplotlib.backends.backend_gtkcairo.py

···

On Mon, 2007-01-15 at 08:26 -0700, Fernando Perez wrote:
---------------------------------------------------------------------------
exceptions.AttributeError Traceback (most
recent call last)

/home/iuser/<ipython console>

/usr/lib/python2.4/site-packages/matplotlib/backends/__init__.py
     52
     53 # a hack to keep old versions of ipython working with mpl
     54 if 'IPython.Shell' in sys.modules:
---> 55 new_figure_manager, draw_if_interactive, show =
pylab_setup()
     56

/usr/lib/python2.4/site-packages/matplotlib/backends/__init__.py in
pylab_setup()
     22 backend_name = 'backend_'+backend.lower()
     23 backend_mod =
__import__('matplotlib.backends.'+backend_name,
---> 24 globals(),locals(),[backend_name])
     25
     26 # Things we pull in from all backends

/usr/lib/python2.4/site-packages/matplotlib/backends/backend_gtkcairo.py
      8
      9 #from matplotlib.backends import backend_cairo
---> 10 import matplotlib.backends.backend_cairo as be_cairo
     11 from matplotlib.backends.backend_gtk import *
     12

AttributeError: 'module' object has no attribute 'backends'

Send instant messages to your online friends http://au.messenger.yahoo.com

All,

I am reviving this old thread because the topic came up in a thread with a less-apropos name, and I think we can move it forward a bit more now. Steve's message (below) is still a useful summary of the cairo status.

examples/backend_driver.py can now be used for wholesale testing and comparison of the cairo backend. Example:

python backend_driver.py cairo.png agg

will make png files using cairo and using agg. On my machine (Thinkpad T60, ubuntu feisty) it takes 0.96 minutes for Cairo, 0.76 for Agg.

There are two conspicuous bugs in the Cairo output:

1) contour_demo.py: the image part of the figure is wrong in shape and content. It looks like an array dimensions confusion or something like that. Strangely, the very similar image in image_demo.py is fine.

2) polar_demo.py: the part of the spiral outside the bounding circle is removed in the Agg version but not in the Cairo version.

There are also subtle differences such as in text positioning. The colorbar in image_masked.py is an example of a case where the agg rendering is much better than the Cairo rendering, which has some sort of aliasing that causes stripes. There are probably other such things worth pointing out.

For pdf output I have not tried to look at all the images, but the timing is Cairo 0.71 minutes, pdf backend 1.56 minutes.

I have also modified matplotlib.use() and the command-line parsing so that the "cairo.pdf" style of backend.output specification works in both places. (Only for cairo--it is not needed for any other backend.)

I am not the right person to try to fix the bugs in the Cairo backend, but I hope it will indeed be moved forward. It looks like it is close, and it could be a big asset for matplotlib in the longer term.

Eric

Steve Chaplin wrote:

···

On Sat, 2007-01-06 at 09:23 -1000, Eric Firing wrote:

Steve,

Darren Dale raised a question offline that I think will be of general interest to mpl devels, and that you are the person to answer:

How do you see the future of a cairo backend as a prospective replacement for most or all of the primary mpl backends?

I think cairo could potentially be used to replace the pdf, ps, svg
and gdk/gtk backends which would unify most of the backends and simplify
a lot of the code.

What would need to be completed in cairo? Based on the cairo web page, I get the impression that quite a bit is still missing, including eps generation and genuine vector ps. But maybe the web page is just out of date.

Which web page is out of date? Where does it mention eps and ps, I
couldn't find it.

My general impression of the cairo "surfaces" is:
ImageSurface/png - support is very good
gtk/xlib - support is very good
ps/pdf/svg are usable but less mature and still developing so there may
            be occasional problems drawing specific items
ps - it used to embed bitmap images but now most output is vector based
eps - is not supported yet, but may be in a future version

What would need to be done in mpl, and how hard would it be?

The cairo backend can already be used for png, ps, pdf and gtk output so
I don't think there would be much to do. Mostly, it needs testing -
running all the mpl examples and checking the output looks OK.

Would mpl get slower if everything went through cairo?

Not sure, you would need to run cairo and test it. It used to be much slower
than Agg but more recent versions have had many optimisations applied and
the difference is much smaller now.

Any other considerations?

Some parts of mpl are Agg-specific and other parts (the whole drawing model)
are designed around the gdk drawing style - this makes things harder and
inefficient when using cairo.

On Sat, 2007-01-06 at 09:36 -1000, Eric Firing wrote:

One more question: how does the image quality of cairo compare to
Agg? Is the antialiasing as good?

Image quality looks OK to me, but I'm no expert.

The web browser Firefox 3.0 (due to be released early in 2007) will use
cairo for all rendering. Firefox requires a high level of graphics
performance and the upcoming cairo 1.4 release is expected to provide
that.

Steve

Send instant messages to your online friends http://au.messenger.yahoo.com

This is a fairly new feature I added to add -- clipping to a polygon.
I haven't ported it to postscript yet. The GraphicsContext now has a
clip_path attribute, which is an agg path storage instance. Even
though it is an agg object (not related to backend_agg actually,
separate code base and wrapping) Eg if we construct a path with

  from matplotlib import agg
  path = agg.path_storage()
  path.move_to(10,10)
  path.line_to(100,100)
  path.line_to(200,200)
  path.line_to(100,200)
  path.close_polygon()

We can access it from python code like
In [62]: for i in range(path.total_vertices()):
   ....: cmd, x, y = path.vertex(i)
   ....: print cmd, x, y
   ....:
1 10.0 10.0
2 100.0 100.0
2 200.0 200.0
2 100.0 200.0
79 0.0 0.0

Where each of the cmd is an integer given by one of the path commands

In [60]: for name in dir(agg):
   ....: if not name.startswith('path_cmd'): continue
   ....: print name, getattr(agg, name)
   ....:
path_cmd_catrom 6
path_cmd_curve3 3
path_cmd_curve4 4
path_cmd_curveN 5
path_cmd_end_poly 15
path_cmd_line_to 2
path_cmd_mask 15
path_cmd_move_to 1
path_cmd_stop 0
path_cmd_ubspline 7

The final command "79" which is the result of the
path.close_polygon() function call is a mask of two flags

In [65]: agg.path_flags_close | agg.path_cmd_end_poly
Out[65]: 79

Steve had expressed some dissatisfaction in using the agg python
object for path storage, because agg is C++ whereas cairo is C, and it
just feels wrong to be using an agg path for cairo, but at the end of
the day, we faced either rolling our own backend independent path
object or reusing one, and since agg is fairly deeply ingrained into
mpl and unlikely to disappear, I decided to do a backend independent
SWIG wrapping of agg that could be used across backends to expose some
of the data structures and functionality (eg image interpolation, path
structures, whatever). The idea was to eventually get rid of backend
agg entirely and replace it with backend_agg2 which uses the SWIG
python layer and thereby exposing more of agg's general functionality
at the python/user layer. Unfortunately, this project has stalled and
we currently have the agg layer but are not using it for agg backend
rendering, which is still done in the custom CXX extension code.

The agg path structure is used very lightly currently at the python
level, so if folks fine this arrangement problematic we can
reconsider.

JDH

···

On 7/6/07, Eric Firing <efiring@...229...> wrote:

2) polar_demo.py: the part of the spiral outside the bounding circle is
removed in the Agg version but not in the Cairo version.