matplotlib-devel Mailing List for matplotlib (Page 2)

>>>>> "Jody" == Jody Winston <jos...@ma...> writes:
 Jody> Any pointers on either finding a Qt backend or writing one?
 Jody> Thanks,
Hi Jody,
Sorry for the delay in getting back to you - I've been out of town for
a week.
There have been several people who have expressed interest in A QT
backend - most recently Ted Drain at the JPL. Perhaps you two could
coordinate your efforts?
The first place to start is in the file
matplotlib/backends/backend_template.py, which serves as a template
for backend writers and gives some instructions. I would follow the
model of one of the *Agg backends, eg TkAgg, GTKAgg, FLTKAgg and
WXAgg, which use Agg to render the image and place it in the GUI
canvas. This is a lot less work and you automiatically get the latest
matplotlib feature set for free. The basic idea is to use a GUI
independent image library that can then be reused across GUIs. 
The backends have to implement concrete versions of several interface
classes: RendererBase, GraphicsContextBase, FigureCanvasBase,
FigureManagerBase, NavigationToolbar2. If you opt to use Agg (or
Cairo) to do the drawing for you, you can leave out RendererBase and
GraphicsContextBase, which are the two that require the most work.
JDH

On the users list, I recently discussed renaming the matlab namespace
to pylab out of trademark concerns
 http://sourceforge.net/mailarchive/message.php?msg_id=10174321
This change is now in CVS. For the next few releases, importing
matplotlib.matlab will work but issue a deprecation warning. I also
added pylab.py to site-packages, and the suggested way of importing
the pylab namespace is
 from pylab import blah, blah
which is a wrapper around
 from matplotlib.pylab import blah, blah
There is a script at http://matplotlib.sf.net/matplotlib_to_pylab.py
to recursively convert names files and directories to the new naming
convention. Read the header to see which cases it handles and which
it does not.
Please update your repositories and test the new changes and
conversion script, which will go into the 0.65 release.
JDH

I've split backend_gtk.py into two parts:
backend_gdk.py - an image backend, like Agg and Cairo.
backend_gtk.py - an GUI backend, that uses gdk rendering
I don't think that a gdk backend is that much use by itself (since Agg
is probably better for most people to use), but I think it has these
benefits:
- the split is more consistent with the way Agg/GTKAgg and
Cairo/GTKCairo are written - the rendering (image backend) is separated
from the GUI.
- it allows me to run examples/backend_driver.py on gdk (without having
gtk widgets popup on screen) and compare performance etc to Agg and
Cairo.
- it makes it easier to delete GDK rendering (while keeping the GTK GUI)
at a later date, if it is no longer required.
Here's the backend_driver results I got for some of the backends,
fastest first:
Backend Template took 0.89 minutes to complete
 template ratio 1.000, template residual 0.000
Backend SVG took 1.06 minutes to complete
 template ratio 1.185, template residual 0.166
Backend PS took 1.32 minutes to complete
 template ratio 1.480, template residual 0.429
Backend Agg took 1.36 minutes to complete
 template ratio 1.519, template residual 0.464
Backend GDK took 1.88 minutes to complete
 template ratio 2.101, template residual 0.984
Backend Cairo took 2.11 minutes to complete
 template ratio 2.358, template residual 1.214
Cairo is slow compared to Agg, but has similar performance to GDK.
This makes sense since GDK and Cairo are both written in Python, 
which much of Agg is written in C++.
Steve

Doing the commands:
------------------------------
from matplotlib.matlab import *
plot((1,2),(3,4))
figtext(0.5,0.5,"$\sqrt 2$")
savefig("tryout.eps")
------------------------------
diplays the sqrt sign shifted upwards against the regular text.
-- 
_________________________________________Norbert Nemec
 Bernhardstr. 2 ... D-93053 Regensburg
 Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
 eMail: <No...@Ne...>

On Dec 7, 2004, at 2:07 AM, Andrew Straw wrote:
> On Dec 6, 2004, at 6:52 PM, Steve Chaplin wrote:
>
>> On Mon, 2004年12月06日 at 15:34 -0500, Perry Greenfield wrote:
>>> I really appreciate Andrew's diagnosing the original problem and
>>> particularly in recognizing it as possibility here. This is a nasty
>>> kind of bug to figure out.
>> The original bug was reported to numarray developers
>
> Probably by the too-modest Steve Chaplin, I suspect. I forgot in my 
> previous email that a significant component of my late-phase debugging 
> consisted of emailing the numarray list, and getting an email from 
> Steven Chaplin, who had independently diagnosed the problem. He had 
> already gone much further than I -- he's the one who submitted the bug 
> report and patch to the glibc itself:
Sorry about that. I should have also thanked Steve for doing the hard 
part.

Hello,
I can confirm that the numarray version of matplotlib
(http://matplotlib.sourceforge.net/) crashes for me without this
patch, and does not crash any more after I applied the patch.
My setup: libc6 version 2.3.2.ds1-18, a Pentium 4 processor (has SSE)
and the atlas3-sse2 library installed.
references:
1) a test program and the patch is at
 http://sources.redhat.com/bugzilla/show_bug.cgi?id=3D10
2) the problem in the context of numarray is described at
 http://sourceforge.net/tracker/index.php?func=3Ddetail&aid=3D870660&gro=
up_id=3D1369&atid=3D450446
I hope this helps,
Jochen
--=20
http://seehuhn.de/

On Tue, 2004年12月07日 at 10:52 +0800, Steve Chaplin wrote:
> On Mon, 2004年12月06日 at 15:34 -0500, Perry Greenfield wrote:
> > I really appreciate Andrew's diagnosing the original problem and 
> > particularly in recognizing it as possibility here. This is a nasty 
> > kind of bug to figure out.
> The original bug was reported to numarray developers via the SourceForge
> bug tracking system back in February, the glibc patch was also applied
> in February. From Numarray 1.0 onwards a 'Special Note' has been
> included in the file numarray/Doc/Install.txt referencing the problem.
> 
> I believe the SourceForge bug report was the one
> 870660 Numarray: CFLAGS build problem 
> yet for some reason I can't locate it anymore. 
Here's the "numarray bugs tracker" link for this report:
http://sourceforge.net/tracker/index.php?func=detail&aid=870660&group_id=1369&atid=450446
My guess is that you were looking in the "numpy bugs tracker" where the
bug was originally filed but which is supposed to be for Numeric:
http://sourceforge.net/tracker/?group_id=1369&atid=101369
Numarray bugs which are filed in the numpy bugs tracker are moved to the
numarray bugs tracker. They're both on the same SF project, but the
numarray tracker is more hidden. I'm sorry this is confusing.
Regards,
Todd

On Dec 6, 2004, at 6:52 PM, Steve Chaplin wrote:
> On Mon, 2004年12月06日 at 15:34 -0500, Perry Greenfield wrote:
>> I really appreciate Andrew's diagnosing the original problem and
>> particularly in recognizing it as possibility here. This is a nasty
>> kind of bug to figure out.
> The original bug was reported to numarray developers
Probably by the too-modest Steve Chaplin, I suspect. I forgot in my 
previous email that a significant component of my late-phase debugging 
consisted of emailing the numarray list, and getting an email from 
Steven Chaplin, who had independently diagnosed the problem. He had 
already gone much further than I -- he's the one who submitted the bug 
report and patch to the glibc itself:
> This is the glibc bug report
> http://sources.redhat.com/bugzilla/show_bug.cgi?id=10
Jochen, that bug report contains a C program which replicates the bug. 
Perhaps you could send that test program to the Debian bug tracking 
system to spur patching? (There is an additional comment on the glibc 
bugzilla page saying "The test program isn't really testing what it is 
supposed to (the SSE status is never touched) but the SSE control 
change is indeed wrong." You may want to address this first if you're 
up for this kind of low-level fun.)
To summarize, we owe a big thanks to Steve Chaplin. A heartfelt 
thanks, Steve!
Cheers!
Andrew

On Mon, 2004年12月06日 at 15:34 -0500, Perry Greenfield wrote:
> I really appreciate Andrew's diagnosing the original problem and 
> particularly in recognizing it as possibility here. This is a nasty 
> kind of bug to figure out.
The original bug was reported to numarray developers via the SourceForge
bug tracking system back in February, the glibc patch was also applied
in February. From Numarray 1.0 onwards a 'Special Note' has been
included in the file numarray/Doc/Install.txt referencing the problem.
I believe the SourceForge bug report was the one
870660 Numarray: CFLAGS build problem 
yet for some reason I can't locate it anymore. Perhaps thats one of the
reasons that the problem keeps getting rediscovered.
This is the glibc bug report
http://sources.redhat.com/bugzilla/show_bug.cgi?id=10
Steve

On Dec 5, 2004, at 9:10 PM, Andrew Straw wrote:
>
> On Dec 5, 2004, at 9:04 AM, Jochen Voss wrote:
>
>> Hello everybody,
>>
>> On Wed, Dec 01, 2004 at 06:38:47PM -0800, Andrew Straw wrote:
>>> It sounds like you may have hit a nasty glibc bug that caused me much
>>> head scratching over the months. Check this thread:
>>>
>>> http://aspn.activestate.com/ASPN/Mail/Message/numpy-discussion/ 
>>> 2207861
>>>
>>> Bottom line:
>>>
>>> numarray does The Right Thing and attempts to set up floating point
>>> exception handling, but older versions of glibc (such as that in 
>>> Debian
>>> sarge) have a bug whereby the floating point error bits in the SSE 
>>> are
>>> not properly cleared, leading to a SIGFPE terminating the program 
>>> the
>>> next time the SSE unit is used.
>> Yes, this was the problem. I applied the patch from
>>
>
I really appreciate Andrew's diagnosing the original problem and 
particularly in recognizing it as possibility here. This is a nasty 
kind of bug to figure out.
>> By the way: I am still interested in my original question. How would
>> I use a debugger etc. to find the problem myself in such a situation?
>
> If you find out, let me know! Seriously, having a process killed by 
> the kernel
Us too!
Perry

Hello Andrew,
On Mon, Dec 06, 2004 at 06:42:03PM +0000, Jochen Voss wrote:
> I will try to add more information to the bug report log.
> Maybe this helps the patch being applied.
It turns out that I do not understand enough of this to
produce an illustrative example. Do you have a minimal
C program which terminates with SIGFPE because of this bug
where it shouldn't?
All the best,
Jochen
--=20
http://seehuhn.de/

Hello Andrew,
On Sun, Dec 05, 2004 at 06:10:47PM -0800, Andrew Straw wrote:
> I have to admit that I'm a little disappointed in the speed of this=20
> bugfix going into the Debian sources. I'd think that since I submitted=
=20
> a 2 line (1 of which was comments) patch over a month ago, which was=20
> copied directly from upstream, this would be about 2 minutes of work=20
> for someone...
Yes, this can be a pain. I think the first thing to do
is to add more information to the bug report log. I guess
the the Debian libc-maintainers are short of time and have
problems to easily see whether the bug is actually a bug and the
fix is actually a fix.
I will try to add more information to the bug report log.
Maybe this helps the patch being applied.
All the best,
Jochen
--=20
http://seehuhn.de/

> By the way: I am still interested in my original question. How would
> I use a debugger etc. to find the problem myself in such a situation?
I should know the answer because I created the glibc patch that fixes
the problem, but it was back in February and I can't remember all the
details.
It started when I thought, why should I run my AthlonXP in '386 emulator
mode' when I can use 'gcc -march=athlon-xp' and actually benefit from
the extra instructions my processor supports. This worked fine until I
compiled numarray and it failed its own tests with a floating-point
exception. But if I used the default gcc settings it worked OK. I filed
a numarray bug report (which I can no longer locate, perhaps they get
deleted after a certain date), they looked at it and said it was
probably a gcc bug. I filed a gcc bug report, and they closed it saying
it was not a gcc bug.
Then I thought it might be a bug with the way kernel handles FP
exceptions and started looking through the kernel sources, but did not
make much progress. So I went back to the numarray source code and tried
no narrow down where the problem was occurring.
Now to answer your question:
Consider you are on a TV game show where you have to guess a number x in
the range 1 to y and are told 'higher', 'lower' or 'correct' after each
turn. You can use a binary search and always guess the mid point of the
range - you are either correct or eliminate half of the possibilities
each turn, so in ceil log(y, 2) turns or less you locate the correct
number.
You can use a similar kind of binary search to locate bugs in software.
You know the bug occurs on some line x of the source code with y lines.
Use gdb and insert breakpoints in the code (I think I just inserted
printf() statements instead of using gdb) and see if the error occurs
before or after the breakpoint, move the breakpoint and try again. The
problem is that source code is rarely a linear list of statements in one
file that are executed in order, but a set of procedures/functions in
many files where the execution order can vary. You can start at the main
() function, split it in half and insert a breakpoint (or printf()) run
it and see in which half the error occurs, repeat the process working
your way down into other functions until you pinpoint the error.
Hope that makes sense. You could now reinstall the old glibc, forget
that you know that glibc is the problem and start again to locate the
bug, it will be useful practise for the next bug that comes along!
Steve

On Dec 5, 2004, at 9:04 AM, Jochen Voss wrote:
> Hello everybody,
>
> On Wed, Dec 01, 2004 at 06:38:47PM -0800, Andrew Straw wrote:
>> It sounds like you may have hit a nasty glibc bug that caused me much
>> head scratching over the months. Check this thread:
>>
>> http://aspn.activestate.com/ASPN/Mail/Message/numpy-discussion/2207861
>>
>> Bottom line:
>>
>> numarray does The Right Thing and attempts to set up floating point
>> exception handling, but older versions of glibc (such as that in 
>> Debian
>> sarge) have a bug whereby the floating point error bits in the SSE are
>> not properly cleared, leading to a SIGFPE terminating the program the
>> next time the SSE unit is used.
> Yes, this was the problem. I applied the patch from
>
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=279294
>
> and the problem disappeared. Does this imply that on current
> Debian/unstable systems matplotlib can only be used with
> python-numeric and not with python-numarray?
You can remove atlas3-sse2 (and perhaps atlas3-sse), and it should run 
OK. Sorry I didn't remember this earlier.
I have to admit that I'm a little disappointed in the speed of this 
bugfix going into the Debian sources. I'd think that since I submitted 
a 2 line (1 of which was comments) patch over a month ago, which was 
copied directly from upstream, this would be about 2 minutes of work 
for someone... Maybe I should have put it at a higher priority than 
"Normal".
> By the way: I am still interested in my original question. How would
> I use a debugger etc. to find the problem myself in such a situation?
If you find out, let me know! Seriously, having a process killed by 
the kernel because of a signal was difficult for me to debug. (Python 
is supposed to insulate you from this kind of low-level stuff and 
generally does a fantastic job.) I had no idea where the FPE signal was 
coming from or why. I first came across this in the context of a 
numarray/Intel IPP program. Because IPP is closed source, I didn't go 
very far initially, and just converted my program to Numeric. Then, I 
encountered the same thing purely within numarray and knew it was 
within my grasp. By making a minimal program that exhibited the bug, I 
determined that the floating point error checking and setting code 
executed on import of numarray.ieeespecial would cause a floating point 
error (SIGFPE) in later matrix calls (e.g. 
numarray.linear_algebra.singular_value_decomposition). I began to 
suspect the SSE unit because this code ran fine when I compiled 
numarray with its built-in lapack_lite.
Strangely, running python from within gdb did not terminate or indicate 
that a SIGFPE was raised. This I'd like to understand a bit better...
Cheers!
Andrew

Hello everybody,
On Wed, Dec 01, 2004 at 06:38:47PM -0800, Andrew Straw wrote:
> It sounds like you may have hit a nasty glibc bug that caused me much=20
> head scratching over the months. Check this thread:
>=20
> http://aspn.activestate.com/ASPN/Mail/Message/numpy-discussion/2207861
>=20
> Bottom line:
>=20
> numarray does The Right Thing and attempts to set up floating point=20
> exception handling, but older versions of glibc (such as that in Debian=
=20
> sarge) have a bug whereby the floating point error bits in the SSE are=20
> not properly cleared, leading to a SIGFPE terminating the program the=20
> next time the SSE unit is used.
Yes, this was the problem. I applied the patch from
 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=3D279294
and the problem disappeared. Does this imply that on current
Debian/unstable systems matplotlib can only be used with
python-numeric and not with python-numarray?
By the way: I am still interested in my original question. How would
I use a debugger etc. to find the problem myself in such a situation?
Many thanks,
Jochen
--=20
http://seehuhn.de/

On Sat, 2004年12月04日 at 19:10 +0100, Norbert Nemec wrote:
> Hi there,
> 
> can anybody reproduce these bugs:
> 
> * when I first use the interactive zoom&move feature and then switch over to 
> the zoom-to-box mode, the mode is not completely switched, but stays in some 
> broken intermediate state: using the mouse, I can draw a box, but at the same 
> time, the plot is moved along with the mouse (like in interactive mode)
I can't reproduce this problem.
> * when I zoom in *very* close, (probably near the limit of floating point 
> resolution), I suddenly see huge garbage polygons in the plot. Probably there 
> should be some check limiting the zoom to the range of floating point values?
I do get this one. It looks like an Agg backend bug because it happens
for me on TkAgg and GTKAgg, but not GTK or GTKCairo.
Steve

Hi there,
can anybody reproduce these bugs:
* when I first use the interactive zoom&move feature and then switch over to 
the zoom-to-box mode, the mode is not completely switched, but stays in some 
broken intermediate state: using the mouse, I can draw a box, but at the same 
time, the plot is moved along with the mouse (like in interactive mode)
* when I zoom in *very* close, (probably near the limit of floating point 
resolution), I suddenly see huge garbage polygons in the plot. Probably there 
should be some check limiting the zoom to the range of floating point values?
Greetings,
Nobbi 
-- 
_________________________________________Norbert Nemec
 Bernhardstr. 2 ... D-93053 Regensburg
 Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
 eMail: <No...@Ne...>

Any pointers on either finding a Qt backend or writing one?
Thanks,
Jody Winston

Jochen Voss wrote:
>I get the following output:
>
> voss@plonk [~/src/mpl/test] ./test.py --numarray 
> fisch
> Floating point exception
>
>Thank you very much,
>Jochen
> 
>
It sounds like you may have hit a nasty glibc bug that caused me much 
head scratching over the months. Check this thread:
http://aspn.activestate.com/ASPN/Mail/Message/numpy-discussion/2207861
Bottom line:
numarray does The Right Thing and attempts to set up floating point 
exception handling, but older versions of glibc (such as that in Debian 
sarge) have a bug whereby the floating point error bits in the SSE are 
not properly cleared, leading to a SIGFPE terminating the program the 
next time the SSE unit is used.
One solution:
Rebuild glibc with the appropriate patch.

>>>>> "Jochen" == Jochen Voss <vo...@se...> writes:
 Jochen> I get the following output:
 Jochen> voss@plonk [~/src/mpl/test] ./test.py --numarray fisch
 Jochen> Floating point exception
I suggest you rm -rf site-packages/matplotlib and
site-packages/numarray and your matplotlib build directory. 
Then do a clean install of both packages, matplotlib cvs and numarray
1.1.1 or cvs. Then try again, being mindful of which backend you are
running under with -dSomeBackend and perhaps increasing the verbose
level.
JDH

Hello Perry,
On Wed, Dec 01, 2004 at 01:28:57PM -0500, Perry Greenfield wrote:
> why not "x =3D sin(t)"? That should work. No need to use map or math.sin
true, this work as such but the crash is still there.
> Does it plot interactively? Do you get any error message or does it
> just crash out of python?
No, it does not plot interactively either.
If I run the script
 from matplotlib.matlab import *
 t=3Darange(0,10,0.1)
 x=3Dsin(t)
 plot(t,x)
 print "fisch"
 show()
I get the following output:
 voss@plonk [~/src/mpl/test] ./test.py --numarray =20
 fisch
 Floating point exception
Thank you very much,
Jochen
--=20
http://seehuhn.de/

Hello,
I tried to use matplotlib with numarray instead of numeric for the
first time, but it seems that it fails for me even for the simplest
examples. For example the script
 from matplotlib.matlab import *
 from math import sin
 t=3Darange(0,10,0.1)
 x=3Dmap(lambda tt:sin(tt), t)
 plot(t,x)
 savefig("out.ps")
fails during the savefig call with a floating point exception.
Unfortunately there is no backtrace printed, so I have no idea where
exactly the problem lies.
Questions: Is there any easy way to get more information about where
the failure happens? Is this specific problem known?
All the best,
Jochen
--=20
http://seehuhn.de/