matplotlib-devel Mailing List for matplotlib (Page 3)

>>>>> "paul" == paul cristini <pau...@un...> writes:
 paul> The pick method because of the need to click on edges did
 paul> not fullfill my needs. So I wrote a new method Called
 paul> PickBigLine that does not required a mouse click close to
 paul> the edge but close to the line you want to pick. This is
 paul> particularly useful after zooming when the edges are
 paul> sometimes out of the axis limits. 
Hi Paul,
It is not clear to me what this method is for. It would help if you
posted an example where the current pick functionality failed and the
one you propose succeeds (perhaps you could define your function at
the top of the file for ease of use).
I have a couple of questions/comments about your code...
 xt, yt = a.get_transform().numerix_x_y(xdata, ydata)
 xt, yt = asarray(xt), asarray(yt) 
There is no need to call asarray since numerix_x_y returns arrays.
 
 xc, yc = xt[1]-xt[0], yt[1]-yt[0]
What is the point of this? Why do you only look at the points xt[1],
xt[0], yt[1], yt[0]? What if someone needs to click on another point
of the line?
 if xc==0.0 and yc == 0.0: return 1000000.
 D = xc*xc + yc*yc
 D1 = -(xt[0]-xywin[0])*yc+(yt[0]-xywin[1])*xc 
 D2 = -(yt[0]-xywin[1])*yc-(xt[0]-xywin[0])*xc
What do D1 and D2 represent? I'm having trouble understanding why,
for example, you need to do (xt[0]-xywin[0])*yc
 if D2/D>1.001 or D2/D<-0.001: return 1000000.
I think the 1000000.0 sentinel value should be renamed to some useful
constant name so it will be self documenting.
 return abs(D1/D)
 artists = self.lines
 if not len(artists): return None
 ds = [ (dist(a),a) for a in artists]
 ds.sort()
 return ds[0][1]
 paul> I also needed to add a
 paul> new property to Line2D called tag (similar to matlab) for
 paul> sorting purposes. I wonder if you have thought of adding
 paul> such a possibility to some objects for which it can be very
 paul> useful.
Does the "label" property help here. Could you give a use case?
Thanks!
JDH

>>>>> "Baptiste" == Baptiste Carvello <bap...@al...> writes:
 Baptiste> While I was at it, I recoded those functions as property
 Baptiste> setters, to put them in line with the rest of
 Baptiste> matplotlib.
Thanks Baptiste - I just committed this to CVS. For future patches,
could you please use mpl naming conventions
 UpperCase - classes
 lower of mixedCase: variables
 lower_underscore : functions
I know this is not prominently documented anywhere, but it will help
keep the code more consistent. I already made the required changes
for your patch.
Thanks again!
JDH

cf...@li... wrote:
>Looks like a problem with drawing ticks. Attached is code that demonstrates the problem (for me), and the traceback.
>
>I'm using Python 2.4.1, Numeric 23.8, and the stock .matplotlibrc, on win32.
>
>The script runs fine in 0.80.
>
>
>[...]
>
>------------------------------------------------------------------------
>
>Traceback (most recent call last):
> File "bug-0.81.py", line 50, in ?
> PlotResults()
> File "bug-0.81.py", line 18, in PlotResults
> cvs.draw()
> File "C:\Python24\Lib\site-packages\matplotlib\backends\backend_agg.py", line 369, in draw
> self.figure.draw(renderer)
> File "C:\Python24\Lib\site-packages\matplotlib\figure.py", line 395, in draw
> for a in self.axes: a.draw(renderer)
> File "c:\Python24\lib\site-packages\matplotlib\axes.py", line 1357, in draw
> self.yaxis.draw(renderer)
> File "C:\Python24\Lib\site-packages\matplotlib\axis.py", line 530, in draw
> self.major.formatter.set_locs(majorLocs)
> File "C:\Python24\Lib\site-packages\matplotlib\ticker.py", line 296, in set_locs
> self._set_orderOfMagnitude(d)
> File "C:\Python24\Lib\site-packages\matplotlib\ticker.py", line 323, in _set_orderOfMagnitude
> else: oom = math.floor(math.log10(locs[-1]))
>OverflowError: math range error
> 
>
Sorry about that. I didn't anticipate the case where only one tick is 
present, and also being equal to zero. The fix has been committed to CVS.
Darren

Hello,
I finally tracked down an annoying ticking bug. The bug goes that way:
1) you plot a figure with many ticks
2) you zoom in so that few ticks are left
3) you call tick_top()
4) you zoom out
Then, some of your ticks will be on top, some others at bottom.
The cause of this bug is that tick_top and friends do not change the visibility 
properties of all the ticks, but only the visible ones. This is because they 
call self.get_major_ticks(), instead of modifying all ticks in self.majorTicks.
While I was at it, I recoded those functions as property setters, to put them in 
line with the rest of matplotlib.
Regards,
BC

Hi everyone,
On Wednesday 08 June 2005 11:42 am, Fernando Perez wrote:
> Darren Dale wrote:
>
> > Would you send me a copies of the bad eps and the fixed version, and al=
so
> > a copy of ps2eps? I don't have that program on my system, and was not
> > able to find it on the web.
>
> Here goes. As it turns out, ps2eps is a simple perl script which I got god
> knows when. So people won't actually have it on their systems, it's
> something that lives in /usr/local/bin on my box, sorry.
>
> But no worries. I can lie with using ps2eps for now, until I can upgrade
> to a moer current ghostscript.
I am still unhappy with the bitmapped fonts that result from the conversion=
 to=20
eps, using ghostscript's epswrite. They look ok in Adobe Reader, but not so=
=20
in kpdf or ggv. (I was inspired to look at this today after discovering tha=
t=20
Adobe Reader for linux does not display correctly when I run in 1024x768=20
resolution on my native 1680x1050 laptop. Kpdf displays correctly, but the=
=20
fonts are terrible.)
I looked into ps2eps some more today, to see if it would generate a file th=
at=20
I could embed in a latex document. The short answer is no.=20
The long answer is that PSFrag was not designed to do what I am trying to d=
o:=20
generate an eps file that can later be embedded in a document. It uses a=20
number of PostScript operators that are illegal in an eps file: setglobal,=
=20
statusdict and userdict. Here is the blurb from PostScript Language=20
Reference, Second Edition, Appendix I: setglobal disrupts page independence=
=20
and nesting of included documents. [...] Creation and modification of globa=
l=20
objects are uneffected by save-restore operators.
I think it might be worth looking into the way PyX is dealing with TeX/LaTe=
X,=20
as someone recently suggested. PyX's eps output looks just like standard (n=
o=20
tex) MPL output, where these nesting issues do not exist. They even have th=
e=20
same issue of dumping entire font definitions into the output.
Darren

 
The pick method because of the need to click on edges did not fullfill
my needs. So I wrote a new method Called PickBigLine that does not
required a mouse click close to the edge but close to the line you want
to pick. This is particularly useful after zooming when the edges are
sometimes out of the axis limits.
I also needed to add a new property to Line2D called tag (similar to
matlab) for sorting purposes. I wonder if you have thought of adding
such a possibility to some objects for which it can be very useful.
Thanks,
Paul Crisini
 def pickBigLine(self, x, y, trans=None):
 """
 Return the Line artist under point that is closest to the x, y. 
if trans
 is None, x, and y are in window coords, 0,0 = lower left. 
Otherwise,
 trans is a matplotlib transform that specifies the coordinate system
 of x, y.
 No need to click on the edge!
 """
 if trans is not None:
 xywin = trans.xy_tup((x,y))
 else:
 xywin = x,y
 def dist(a):
 xdata = a.get_xdata()
 ydata = a.get_ydata()
 xt, yt = a.get_transform().numerix_x_y(xdata, ydata)
 xt, yt = asarray(xt), asarray(yt)
 xc, yc = xt[1]-xt[0], yt[1]-yt[0]
 if xc==0.0 and yc == 0.0: return 1000000.
 D = xc*xc + yc*yc
 D1 = -(xt[0]-xywin[0])*yc+(yt[0]-xywin[1])*xc
 D2 = -(yt[0]-xywin[1])*yc-(xt[0]-xywin[0])*xc
 if D2/D>1.001 or D2/D<-0.001: return 1000000.
 return abs(D1/D)
 artists = self.lines
 if not len(artists): return None
 ds = [ (dist(a),a) for a in artists]
 ds.sort()
 return ds[0][1]

Looks like a problem with drawing ticks. Attached is code that demonstrates the problem (for me), and the traceback.
I'm using Python 2.4.1, Numeric 23.8, and the stock .matplotlibrc, on win32.
The script runs fine in 0.80.
Cheers,
Chris Fuller

John Hunter wrote:
> TeX support : Now you can (optionally) use TeX to handle *all* of the
> text elements in your figure with the rc param text.usetex (*Agg and
> PS only). PS support requires tex, dvips and Ghostscript 8.51
> (older versions do not work properly -- test your version with
> 'gs --version'). Agg support requires tex and dvipng. A
> directory ~/.tex.cache is created where support files are cached for
> later reuse. We opted to ues TeX rather than LaTeX because it is
> faster and can do all the things we thought useful for figure text
> snippets. See http://matplotlib.sf.net/screenshots.html#tex_demo
> and http://matplotlib.sf.net/matplotlib.texmanager.html. There are
> several new rc params for configuring tex/latex support
Yes! Many, many thanks to all who helped on this. Below are some notes 
for Fedora users, who may find it a bit tricky to get this thing working.
The new mpl TeX support requires dvipng, which is not available in any 
Yum repository I could find. It seems that FedoraCore 4 will include 
tetex-3.0, which bundles dvipng. But in Fedora Core 2-3, it is not 
available. You can obviously build it from sources from:
http://sourceforge.net/project/showfiles.php?group_id=109426&package_id=118693&release_id=302829
but if like me, you are lazy and like to keep all your machines 
centrally administered thanks to the convenience of Yum, you can do 
better. Here, I found a dvipng rpm built for Fedora 2, but which works 
fine in FC3:
http://rpm.nogin.org/WebWork/fc2/dvipng-1.4-1.rhfc2.i386.html
Just drop that RPM somewhere in your local Yum repo (or just install it 
by hand), and you're done. TeX beauty in all your mpl plots!
Again, eternal gratitude to Darren, John and the rest.
Cheers,
f

Win32 Warning: This is the first release I've done since my windows
 build syste dies and I had to reinstall a bunch of tools and some
 version numbers in my GTK build environment changed (eg my GTK
 setup). Let me know if you encounter any problems. As always, try
 removing site-packages/matplotlib and reinstalling before reporting
 any problems
TeX support : Now you can (optionally) use TeX to handle *all* of the
 text elements in your figure with the rc param text.usetex (*Agg and
 PS only). PS support requires tex, dvips and Ghostscript 8.51
 (older versions do not work properly -- test your version with
 'gs --version'). Agg support requires tex and dvipng. A
 directory ~/.tex.cache is created where support files are cached for
 later reuse. We opted to ues TeX rather than LaTeX because it is
 faster and can do all the things we thought useful for figure text
 snippets. See http://matplotlib.sf.net/screenshots.html#tex_demo
 and http://matplotlib.sf.net/matplotlib.texmanager.html. There are
 several new rc params for configuring tex/latex support
 # use tex/latex for all text handling
 text.usetex : False 
 # tex is faster, but latex is required to use special font
 # packages. See font.latex.package
 text.tex.engine : latex 
 
 # This must be an available LaTeX font package, 
 # like 'times' or 'pslatex' ; only applies if text.usetex 
 # is true
 font.latex.package : type1cm 
 Special thanks to Darren Dale for lots of hair-pulling work
 customizing, enhancing and debugging the ps backend for LaTeX
 support.
Masked arrays: Support for masked arrays in line plots, pcolor and
 contours. There are some problems with filled contours and masked
 arrays. Thanks Eric Firing and Jeffrey Whitaker.
Contour levels arg changes: see http://matplotlib.sf.net/API_CHANGES
 for details
Byte images: Much faster imaeg loading for MxNx4 or MxNx3 UInt8
 images, which bypasses the memory and CPU intensive integer/floating
 point conversions. Nicolas Girard
New image resize options interpolation options. New values for the
 interp kwarg are
 
 'nearest', 'bilinear', 'bicubic', 'spline16', 'spline36',
 'hanning', 'hamming', 'hermite', 'kaiser', 'quadric', 'catrom',
 'gaussian', 'bessel', 'mitchell', 'sinc', 'lanczos', 'blackman'
 See help(imshow) for details, particularly the interpolation,
 filternorm and filterrad kwargs.
Text and dashes - Daishi Harada contributed a patch for connecting
 text to points with lines. See examples/dashpointlabel.py and
 examples/dashtick.py
Fast markers on win32: The marker cache optimization is finally
 available for win32, after an agg bug was found and fixed (thanks
 Maxim!). Line marker plots should be considerably faster now on
 win32.
set deprecated: use setp instead; a simple, mostly braindead,
 conversion script is provided below
Qt in ipython/pylab: You can now use qt in ipython pylab mode. Thanks
 Fernando Perez and the Orsay team!
Agg wrapper proper: Started work on a proper agg wrapper to expose
 more general agg functionality in mpl. See examples/agg_test.py.
 Lots of wrapping remains to be done.
New scalar formatter: Darren Dale did a lot of work to make scalar
 formatting smarter in pathalogical cases. See
 examples/newscalarformatter_demo.py
Small features: linewidth and faceted kwarg to scatter to control
 edgewidth and color, autolegending now inspects line segments in
 addition to vertices, upgraded to agg23, new example showing how to
 use line collections examples/line_collection.py, fixed antialiased
 property setting in agg, added a postscript papersize rc option,
 added an example showing how to embed mpl in a qt app
 examples/embedding_in_qt.py, arrow keys now exposed in mpl's GUI
 neutral event handling, added "among" kwarg to axes picker function
 to limit picks, added autoscale_on property to Axes to control
 whether or not autoscaling is done.
Bug fixes: fixed a contour masked array bug, contour memory leak
# Here is a script to recursively convert set and get to setp and
# getp. Please backup entire directory recursively before
# running this script
from matplotlib.cbook import listFiles
for fname in listFiles('.', '*.py'):
 lines = []
 cnt = 0
 for line in file(fname):
 if line.lstrip().startswith('set('):
 line = line.replace('set(', 'setp(')
 cnt +=1
 if line.lstrip().startswith('get('):
 line = line.replace('get(', 'getp(')
 cnt +=1
 lines.append(line)
 
 file(fname, 'w').writelines(lines)
 print '%s\t: %d replacements'%(fname,cnt)

>>>>> "Stefan" == Stefan Kuzminski <pon...@ya...> writes:
 Stefan> Is there an easy way to hide the axis lines? ( bounding
 Stefan> box of the figure ). I set the xaxis._visible to false
 Stefan> and that hid the ticks, I also tried setting figure(
 Stefan> edgecolor='white' ).
This may be a confusion of terminology.
In matplotlib, the figure box is the entire canvas, and is
controlled by fig.figurePatch. The Axes bounding box is the white box
inside the figure box, and is controlled by ax.axesPatch. Both are
matplotlib.patches.Rectangle instances. The Axis objects are the X
and Y Axis and control the ticks and tick labels.
To turn off the Axes box entirely, you can do
 ax.set_frame_on(False)
JDH

Is there an easy way to hide the axis lines? ( bounding box of the
figure ). I set the xaxis._visible to false and that hid the ticks, I
also tried setting figure( edgecolor='white' ).
thanks,
S 
--- John Hunter <jdh...@ac...> wrote:
> >>>>> "Stefan" == Stefan Kuzminski <pon...@ya...> writes:
> 
> Stefan> I'm wondering how difficult it would be to add a
> Stefan> speedometer-like 'dial' plot type. ( apparently this
> Stefan> appeals to executive types ) I imagine it would take a
> Stefan> range and a scalar value which would be the location of
> Stefan> the 'needle'. Any pointers to how to craft such a thing
> Stefan> would be appreciated.
> 
> I think you could reuse most of the PolarAxes code for this, but
> instead of using a polar transform place a line or line/arrow in the
> center pointing to the "speed"
> 
> JDH
> 
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

Hi all,
	We have updated the OSX binaries which include all dependencies in a
single installer (i.e. libpng, freetype, numarray, Numeric, ...). There
are 10.3/panther and 10.4/tiger binaries which include mpl-0.80 and
basemap-0.4.2. The panther binary includes wxpython and the tiger
binary uses the system included wxpython. Please feel free to contact
me if you experience any problems.
"Tools, Applications" Section:
http://sda.iu.edu/projects.html
- Charlie

On Mon, 2005年06月06日 at 08:58 -0500, John Hunter wrote:
> >>>>> "Nicholas" == Nicholas Young <su...@su...> writes:
> 
> Nicholas> Hi, There have been several times when I've come across
> Nicholas> conditions where I want a user to be able to pick from
> Nicholas> amongst a limited number of objects on plots containing
> Nicholas> lots of other objects. Previously I've written my own
> Nicholas> functions to do this but it occurs to me that this is
> Nicholas> likely to be something others will want to do. To this
> Nicholas> end I've made a very simple addition to the Axes.pick
> Nicholas> method to take a keyword argument test which, if a
> Nicholas> callable, is used to filter the objects from which to
> Nicholas> pick. A patch to CVS containing the alteration is
> Nicholas> enclosed.
> 
> Hi Nick,
> 
> Although using a function may be the most general way of making
> picking selective, I wonder if it is the most intuitive. I suspect
> that 9 time out of 10, a user will just want to pass a list of objects
> on which to pick. Do you agree?
I do - but for flexibility how about this patch which allows both
options? The kind of list intersection I've used isn't the most
efficient but as mpl isn't at its most efficient with large numbers of
artists I didn't think that would be a problem.
Nick

>>>>> "Nicholas" == Nicholas Young <su...@su...> writes:
 Nicholas> Hi, There have been several times when I've come across
 Nicholas> conditions where I want a user to be able to pick from
 Nicholas> amongst a limited number of objects on plots containing
 Nicholas> lots of other objects. Previously I've written my own
 Nicholas> functions to do this but it occurs to me that this is
 Nicholas> likely to be something others will want to do. To this
 Nicholas> end I've made a very simple addition to the Axes.pick
 Nicholas> method to take a keyword argument test which, if a
 Nicholas> callable, is used to filter the objects from which to
 Nicholas> pick. A patch to CVS containing the alteration is
 Nicholas> enclosed.
Hi Nick,
Although using a function may be the most general way of making
picking selective, I wonder if it is the most intuitive. I suspect
that 9 time out of 10, a user will just want to pass a list of objects
on which to pick. Do you agree?
JDH

Hi,
There have been several times when I've come across conditions where I
want a user to be able to pick from amongst a limited number of objects
on plots containing lots of other objects. Previously I've written my
own functions to do this but it occurs to me that this is likely to be
something others will want to do. To this end I've made a very simple
addition to the Axes.pick method to take a keyword argument test which,
if a callable, is used to filter the objects from which to pick. A
patch to CVS containing the alteration is enclosed.
Nick

I worked quite a bit on the TeX support in MPL this weekend. I think the co=
de=20
is ready for wider testing, the changes are in CVS. Note there are a few ne=
w=20
options in .matplotlibrc:=20
1) font.latex.package allows you to select the latex fonts, but your select=
ion=20
has to be a valid latex package.
2) text.tex.engine allows you to select tex or latex to handle the text=20
layout.
3) ps.usedistiller allows users to pass any ps/eps output through=20
ghostscript's distiller (this is a necessary step for creating eps output i=
f=20
tex is handling the text). This will convert the fonts, shrink the file siz=
e,=20
and the resulting file should open more quickly. I have tested with=20
gnu-ghostscript-8.16. ghostscript-7.07 seems to have trouble getting the ep=
s=20
bounding box right.
4) ps.distiller.res allows the dpi to be set. It defaults to 600, but I=20
personally have it set to 6000 to get the best on-screen results for=20
files embedded in tex and converted to pdf. (JDH- note this option was=20
briefly called text.tex.epsres)
I also silenced the tex, dvips and gs stdout messages. If you want to see a=
ll=20
the reports that they print, you can set verbose.level =3D debug-annoying
Changes are in CVS.
Darren

>>>>> "Stefan" == Stefan Kuzminski <pon...@ya...> writes:
 Stefan> I'm wondering how difficult it would be to add a
 Stefan> speedometer-like 'dial' plot type. ( apparently this
 Stefan> appeals to executive types ) I imagine it would take a
 Stefan> range and a scalar value which would be the location of
 Stefan> the 'needle'. Any pointers to how to craft such a thing
 Stefan> would be appreciated.
I think you could reuse most of the PolarAxes code for this, but
instead of using a polar transform place a line or line/arrow in the
center pointing to the "speed"
JDH

I'm wondering how difficult it would be to add a speedometer-like
'dial' plot type. ( apparently this appeals to executive types ) I
imagine it would take a range and a scalar value which would be the
location of the 'needle'. Any pointers to how to craft such a thing
would be appreciated.
thanks,
Stefan
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

John Hunter wrote:
>>>>>>"Jeff" == Jeff Whitaker <js...@fa...> writes:
>>>>>> 
>>>>>>
>
> Jeff> Eric: The combination of your two patches (cntr_bugfix.diff
> Jeff> and cntr_bugfix2.diff) does indeed fix all of the problems
> Jeff> with the basemap examples. Many thanks!
>
>Are you testing from CVS ? I committed them and want to make sure I
>have the right version in.
>
>Thanks to all...
>
>JDH
> 
>
John: Following up on that 'ZeroDivisionError' - it goes away when I 
replace ticker.py with version 1.26 from CVS. Must be due to Darren's 
changes yesterday.
To answer your original question - yes, it appeare you committed all the 
necessary fixes to cntr.c.
-Jeff
-- 
Jeffrey S. Whitaker Phone : (303)497-6313
NOAA/OAR/CDC R/CDC1 FAX : (303)497-6449
325 Broadway Web : http://www.cdc.noaa.gov/~jsw
Boulder, CO, USA 80305-3328 Office: Skaggs Research Cntr 1D-124

John Hunter wrote:
>>>>>>"Jeff" == Jeff Whitaker <js...@fa...> writes:
>>>>>> 
>>>>>>
>
> Jeff> Eric: The combination of your two patches (cntr_bugfix.diff
> Jeff> and cntr_bugfix2.diff) does indeed fix all of the problems
> Jeff> with the basemap examples. Many thanks!
>
>Are you testing from CVS ? I committed them and want to make sure I
>have the right version in.
>
>Thanks to all...
>
>JDH
> 
>
John: I was using a cvs version from the morning of 6/1, with Eric's 
two patches applied. I just tried a new cvs checkout (from the evening 
of 6/2), but now whatever script I run I get the 'Divide by Zero' 
errors reported by Nicholas Girard yesterday.
-Jeff
-- 
Jeffrey S. Whitaker Phone : (303)497-6313
NOAA/OAR/CDC R/CDC1 FAX : (303)497-6449
325 Broadway Web : http://www.cdc.noaa.gov/~jsw
Boulder, CO, USA 80305-3328 Office: Skaggs Research Cntr 1D-124

>>>>> "Jeff" == Jeff Whitaker <js...@fa...> writes:
 Jeff> Eric: The combination of your two patches (cntr_bugfix.diff
 Jeff> and cntr_bugfix2.diff) does indeed fix all of the problems
 Jeff> with the basemap examples. Many thanks!
Are you testing from CVS ? I committed them and want to make sure I
have the right version in.
Thanks to all...
JDH

Eric Firing wrote:
> John, Jeff,
>
> The attached patch against the cntr.c in cvs does the following:
>
> 1) Fixes the bug you (Jeff) described above.
>
> 2) Adds a print function for debugging, but normally unused.
>
> 3) Fixes a minor bug in which space for an array of ints was allocated 
> where only an array of chars was used.
>
> The original masked array contourf bug is still there, but I suspect 
> (and hope!) it will not present a serious problem for basemap. I 
> still want to find and fix it, but that may take a long time.
>
> I want to make some minor cleanups and one API change in contour.py, 
> but I will leave that for a separate message, probably within 24 hours.
>
> Eric
Eric: The combination of your two patches (cntr_bugfix.diff and cntr_bugfix2.diff) does indeed fix all of the problems with the basemap examples. Many thanks!
-Jeff
-- 
Jeffrey S. Whitaker Phone : (303)497-6313
Meteorologist FAX : (303)497-6449
NOAA/OAR/CDC R/CDC1 Email : Jef...@no...
325 Broadway Office : Skaggs Research Cntr 1D-124
Boulder, CO, USA 80303-3328 Web : http://tinyurl.com/5telg

On Thursday 02 June 2005 1:55 pm, you wrote:
> Darren Dale <dd...@co...> writes:
> [snip]
>
> > the ps backend will produce a transitional eps file with a bunch of
> > psfrag markers, along with an associated tex file. backend_ps runs latex
> > on the tex file, which embeds the eps, and replaces the markers with the
> > actual text. Now we have a ps file, which can be converted into an eps =
if
> > desired with ps2epsi, or maybe this could be changed to a call to
> > ghostscript: gs -dNOPAUSE -sDEVICE=3Depswrite -sOutputFile=3Dtex_demo.e=
ps
> > tex_demo.ps. As far as the user is concerned, you type savefig('foo.eps=
')
> > and get foo.eps, blissfully unaware of all the intermediate steps.
>
> http://www.tex.ac.uk/cgi-bin/texfaq2html?label=3Dmpprologues
>
> describes a similar situation, but they use dvips -E to generate the
> eps file directly. Could you do that rather than using ghostscript?
> Apart from anything else, it should be quicker.
I had tried this. -E does not seem to be very robust, I cant view the=20
resulting eps files, they appear to be corrupted.
>
> > Here is my current problem. I embed an eps file generated with ps2epsi =
in
> > a new tex document. The dvi looks ok, but after dvips, the text in my
> > figure gets inverted and shifted. This seems to be a bigger problem when
> > using latex classes that do some font handling behind the scenes, for
> > example, article seems ok, but revtex4 and ha-prosper yield unacceptable
> > results.
>
> As an alternative, could you use the .tex/.eps pair directly. This
> would allow you to use the same font and text size in the figure
> caption as in the main document.
This is the way it was originally done. I think it is much better to have a=
=20
single nugget than to have a half complete figure. I think scientific=20
journals may have a problem with submitting these kinds of half complete=20
figures for publication.
>
> > On the other hand, if I use ghostscript to generate the eps, I can
> > embed the files and the printed page looks fine, but the fonts look
> > terrible on screen.
>
> If you are using xdvi, then I found that too. I blamed it on
> ghostscript being used to render the graphics with low resolution, or
> not antialiasing the text.
>
[...]
> >
> > How can MPL deal with these non-MPL issues? I know John doesn't like the
> > idea of handling text in eps as images, but if we did, could that be a
> > suitable workaround?
>
> I agree with John here.
So do I. But I have a thesis, two papers, and two talks to write. I was hop=
ing=20
practicality might temporarily win out, but I dont know for sure that such =
an=20
effort would even solve the problem.
>
> > Does anyone have another suggestion?
>
> Hope these suggestions might help.
>
> Chris
>
> PS One further thought I had is MetaPost.
> http://www.tex.ac.uk/cgi-bin/texfaq2html?label=3DMP Is it possible for
> matplotlib to embed the tex expressions in the same way that metapost
> does? This might not solve your problem however, as xdvi currently has
> a problem displaying them - see
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=3D295205 for details.
>
> PPS If you need something in a hurry (for your thesis?), can you
> create the whole thing as a bitmap? or use the latex "ite" package to
> position labels? If you do the latter, there is a patch I should send
> you - otherwise it clashes with various packages.
Thank you for the suggestions, but they are both significant deviations fro=
m=20
what I already have in place, and I have to prepare a talk for June 13. In=
=20
the worst case scenario, I will just set usetex to false, and make some=20
temporary changes in my local installation to work around some font layout=
=20
issues that usetex had addressed.
=46onts, fonts, fonts. My new f-word.
=2D-=20
Darren S. Dale
Bard Hall
Department of Materials Science and Engineering
Cornell University
Ithaca, NY. 14850
dd...@co...
http://people.ccmr.cornell.edu/~dd55/

Nicolas Girard wrote:
> That beeing said, I'm also in the need of a way of saving eps figures of gray 
> contours in negative, in order to save toner... with the standard gray 
> colormap, my contours are very dark, leading to a huge consumption of toner.
> 
> With a gray colormap "in negative", the result would be at least as readable 
> and consumes much less toner.
> 
> Two strategies are possible here:
> 
> - either display the gray colormap as usual on the screen, but save the figure 
> in negative with the "eps" or "png" backends (this is AFAIK the default 
> behaviour of IDL),
-1. One of mpl's strengths is that it shows you on screen what you get 
on paper, I'd _hate_ to folow IDL here (or for that matter, on just 
about anything else).
> - or provide a "negative gray" colormap ; I've tried to hack the cm.py in that 
> way, and tried something like:
> 
> _ngray_data = {'red': ((0., 1, 1), (1., 0, 0)),
> 'green': ((0., 1, 1), (1., 0, 0)),
> 'blue': ((0., 1, 1), (1., 0, 0))}
I've meen meaning to add a .reversed() method to the colormap instances, 
which will automatically return an instance of the given colormap with 
its lookup table reversed, and the name simply with a suffix appended 
(_r, _rev, r, ...). I think this is the right solution, and I also 
think we should supply, by default, reversed versions of the standard 
colormaps. At the very least, the gray_rev (or whatever it gets called) 
can be _extremely_ useful.
> but there seems to be a problem with the highest values, that keep beeing 
> displayed in white instead of black.
Mmh, I haven't seen this. I have the same defined in my default 
run-time config loaded by ipython, and I haven't seen this kind of 
clipping, but maybe I just haven't noticed it. I've been using 
matshow() and haven't seen this, I'll look again in more detail later.
Cheers,
f

Darren Dale <dd...@co...> writes:
[snip]
> the ps backend will produce a transitional eps file with a bunch of psfrag 
> markers, along with an associated tex file. backend_ps runs latex on the tex 
> file, which embeds the eps, and replaces the markers with the actual text. 
> Now we have a ps file, which can be converted into an eps if desired with 
> ps2epsi, or maybe this could be changed to a call to ghostscript: gs 
> -dNOPAUSE -sDEVICE=epswrite -sOutputFile=tex_demo.eps tex_demo.ps. As far as 
> the user is concerned, you type savefig('foo.eps') and get foo.eps, 
> blissfully unaware of all the intermediate steps.
http://www.tex.ac.uk/cgi-bin/texfaq2html?label=mpprologues
describes a similar situation, but they use dvips -E to generate the
eps file directly. Could you do that rather than using ghostscript?
Apart from anything else, it should be quicker.
> 
> Here is my current problem. I embed an eps file generated with ps2epsi in a 
> new tex document. The dvi looks ok, but after dvips, the text in my figure 
> gets inverted and shifted. This seems to be a bigger problem when using latex 
> classes that do some font handling behind the scenes, for example, article 
> seems ok, but revtex4 and ha-prosper yield unacceptable results.
As an alternative, could you use the .tex/.eps pair directly. This
would allow you to use the same font and text size in the figure
caption as in the main document. 
> 
> On the other hand, if I use ghostscript to generate the eps, I can
> embed the files and the printed page looks fine, but the fonts look
> terrible on screen. 
If you are using xdvi, then I found that too. I blamed it on
ghostscript being used to render the graphics with low resolution, or
not antialiasing the text.
> Here is a comment from the afpl website 
> http://www.cs.wisc.edu/~ghost/doc/AFPL/8.50/Issues.htm:
> 
> ---
> pswrite device generates low level PostScript.
> 
> pswrite and epswrite devices reduce everything to path, fill, stroke, clip 
> image, and imagemask operations. Although the resulting file prints OK it 
> produces unsatisfactory results when scaled, distilled or imported into 
> graphic editors. The file can easily exceed 4GB and hit file size limits in 
> some applications or operation systems. Handling of big files is slow. 
> ---
> The associated bug report (filed over 2 years ago) says this is a low priority 
> problem.
> 
> How can MPL deal with these non-MPL issues? I know John doesn't like the idea 
> of handling text in eps as images, but if we did, could that be a suitable 
> workaround? 
I agree with John here. 
> Does anyone have another suggestion?
Hope these suggestions might help. 
Chris
PS One further thought I had is MetaPost.
http://www.tex.ac.uk/cgi-bin/texfaq2html?label=MP Is it possible for
matplotlib to embed the tex expressions in the same way that metapost
does? This might not solve your problem however, as xdvi currently has
a problem displaying them - see
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=295205 for details.
PPS If you need something in a hurry (for your thesis?), can you
create the whole thing as a bitmap? or use the latex "ite" package to
position labels? If you do the latter, there is a patch I should send
you - otherwise it clashes with various packages.