matplotlib-users Mailing List for matplotlib

Hi Sterling ,
Thanks for your email. I definitely think I was running into issues with
the figure updating while it was trying to draw , constantly. I
experimented with sleep ..but didnt try hard enough to get it to work.
That said, I have a very nice solution to my problem using the
wx.aui.AuiNotebook
 borrowed from example wx5 (
http://matplotlib.org/examples/user_interfaces/embedding_in_wx5.html)
# MY SOLUTION
I now have a single Plot wx.Panel which then encloses a wx.aui.Notebook
just like in the wx5 example link above.
Once I have finished processing the data . I initialize the plotter just
like in the example above. I have a for loop that goes over all 384 data
sets and then adds a page to the Notebook for every dataset.
The result is a 384 , tabbed Notebook , where each tab is a fully
interactive matplotlib plot , created like before.
def do_my_plot(well_id, plotter_from_main):
 my_plotter = plotter_from_main
 # Here is where I add a page to the Notebook and get its current axis
 ax = plotter.add("figure %s" % well_id).gca()
 par1 = ax .twinx()
 par2 = ax.twinx()
.............................................and -so on
The neatest thing of the final Notebook is that , although it takes about a
minute to appear at first call to frame.Show()..
After that it is very performant.
 Importantly I can quickly scroll through all the 384 tabs ( or pages in
the notebook) using CTRL-TAB to go forward and CTRL-Shift_TAB to go
backward. This ability to navigate is a huge plus for me. The Navigation is
actually very fast and "redraws" are nearly instantaneous.
I will still try and get the "single figure that refreshed " approach as I
originally wanted, using the techniques you , Ben and Damon suggested .
In many cases I dont need to have a 384 tabbed frame..and can do with a
plot that refreshes every few seconds so I know that everything went right.
Thanks everyone for all your help..Ill get back to the group once I get the
repaint within a single frame to work..
Hari
On Wed, Oct 17, 2012 at 5:38 PM, Sterling Smith <sm...@fu...>wrote:
> Hari,
>
> While I am not intimately acquainted with the inner working of the
> interactive matplotlib functionality, I have seen that it tries to not
> update the figure if you ask for some change to it while it is trying to
> update the figure. That sounds circular, but oh well.
>
> Perhaps you could have each analysis open a new figure, and have an if
> statement to close 5 (or 10...) figures ago.
>
> Another subtlety that I have noticed (and perhaps read somewhere) is that
> there could be a difference in behavior between having interactivity set in
> the matplotlibrc file and using the ion() call after having set
> interactive: False in the matplotlibrc file.
>
> Another solution might be a time.sleep after each update of the figure.
>
> (Note that with ion(), the command for updating the figure is pylab.draw,
> which may need to be issued after each case - the pylab/pyplot functions
> usually have a draw_if_interactive call in them.)
>
> -Sterling
>
> PS If I am causing more confusion than help, please let me know.
>
>
> On Oct 17, 2012, at 10:54AM, hari jayaram wrote:
>
> > Thanks Benjamin, Sterling and Damon for your prompt help
> >
> > However I am still not able to achieve what I wanted .
> >
> > I can get the headless script to work just great where it saves all the
> figures and I can view them after the script is done running.
> >
> > But somehow when I try the figure number method that Sterling suggested
> , along with the axis clear and redraw method (Damon) , or the decouple and
> clear and then plot method (Benjamin Root) : I get the plot just spinning
> with a blue circle on Windows 7 and the script just chugs merrily along.
> >
> >
> > I think part of the problem was that I was wrong in the way I stated my
> application. Each of the 384 data processing steps takes a few seconds..and
> not a minute as I had indicated. I tried with both ion() and ioff() and
> giving the figure a number , which stays constant and clearing the axis
> everytime before plotting. But I get a spiining blue circle in Windows.
> >
> > I will try and cookup a test case , and send to the list , to reproduce
> what I am seeing. it may still be that I am calling pylab , pyplot
> incorrectly and hence not getting the continuously changing figure that
> your suggestions should give me.
> >
> > hari
> >
> >
> >
> >
> > Using plt.ion() or plt.ioff() causes a spinning blue-ball on
> windows..while the rest of the script continues.
> > If I use the figure number trick. I get the first figure displayed.
> >
> > On Tue, Oct 16, 2012 at 12:15 PM, Benjamin Root <ben...@ou...> wrote:
> >
> >
> > On Tue, Oct 16, 2012 at 11:25 AM, hari jayaram <ha...@gm...>
> wrote:
> > Hi
> > I am a relative newbie to matplotlib.
> >
> > I have a python script that handles a dataset that comprises 384 sets of
> data.
> >
> > At the present moment , I read in a set of data - process it - and the
> create a figure using code shown below.
> > I am using windows with the default backend ( I think I set it to wx).
> >
> > When I run the program, figure after figure shows up..the program
> continues from well to well plotting the figure. I can close the figure
> window using the X on the right -hand side..while the program chugs along.
> >
> > Is there a way to just recycle the figure object , so that the plot
> shows up for a brief second and refreshes when the next calculation is
> complete. Each process_data function , takes a few minutes.
> >
> > Alternatively I just want to close the figure object I show after a
> brief lag. I am OK if that happens instantaneously..but I dont know how to
> achieve this.
> > Do I have to use the matplotlib.Figure object to achieve this
> functionality
> >
> > Thanks
> > Hari
> >
> >
> >
> > Hari,
> >
> > To recycle the figure, try the following:
> >
> >
> >
> > import matplotlib.pyplot as plt
> >
> > def do_my_plot(par1, par2, well_id):
> > processed_data_object = processed_dict[well_id]
> > # Plot all the data
> > par1.plot(processed_data_object.raw_x,processed_data_object.raw_y).
> > par2.plot(....
> > # finally
> > plt.show()
> > # I tried fig.clf()
> >
> >
> > def plot_and_process_data():
> > plt.ion() # Turn on interactive mode
> > fig = plt.figure(figsize=(7,7)
> > ax = fig.add_subplot(1,1,1)
> > par1 =ax.twinx()
> > par2 = ax.twinx()
> >
> > for well_id in list_of_384_well_ids:
> > par1.cla()
> > par2.cla()
> > process_data(well_id)
> > do_my_plot(par1, par2, well_id)
> >
> > Note, this is completely untested, but it would be how I would go about
> it at first. The "plt.ion()" turns on interactive mode to allow your code
> to continue running even after the plot window appears (but does not end
> until the last window is closed.). Of course, another approach would
> simply be to do "fig.savefig()" after every update to the figure and never
> use show() and ion() (essentially, a non-interactive head-less script).
> >
> > Hopefully, this helps.
> > Ben Root
> >
> >
> >
> ------------------------------------------------------------------------------
> > Everyone hates slow websites. So do we.
> > Make your web apps faster with AppDynamics
> > Download AppDynamics Lite for free today:
> >
> http://p.sf.net/sfu/appdyn_sfd2d_oct_______________________________________________
> > Matplotlib-users mailing list
> > Mat...@li...
> > https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>

Hari,
While I am not intimately acquainted with the inner working of the interactive matplotlib functionality, I have seen that it tries to not update the figure if you ask for some change to it while it is trying to update the figure. That sounds circular, but oh well. 
Perhaps you could have each analysis open a new figure, and have an if statement to close 5 (or 10...) figures ago. 
Another subtlety that I have noticed (and perhaps read somewhere) is that there could be a difference in behavior between having interactivity set in the matplotlibrc file and using the ion() call after having set interactive: False in the matplotlibrc file.
Another solution might be a time.sleep after each update of the figure.
(Note that with ion(), the command for updating the figure is pylab.draw, which may need to be issued after each case - the pylab/pyplot functions usually have a draw_if_interactive call in them.)
-Sterling
PS If I am causing more confusion than help, please let me know.
On Oct 17, 2012, at 10:54AM, hari jayaram wrote:
> Thanks Benjamin, Sterling and Damon for your prompt help
> 
> However I am still not able to achieve what I wanted .
> 
> I can get the headless script to work just great where it saves all the figures and I can view them after the script is done running.
> 
> But somehow when I try the figure number method that Sterling suggested , along with the axis clear and redraw method (Damon) , or the decouple and clear and then plot method (Benjamin Root) : I get the plot just spinning with a blue circle on Windows 7 and the script just chugs merrily along.
> 
> 
> I think part of the problem was that I was wrong in the way I stated my application. Each of the 384 data processing steps takes a few seconds..and not a minute as I had indicated. I tried with both ion() and ioff() and giving the figure a number , which stays constant and clearing the axis everytime before plotting. But I get a spiining blue circle in Windows.
> 
> I will try and cookup a test case , and send to the list , to reproduce what I am seeing. it may still be that I am calling pylab , pyplot incorrectly and hence not getting the continuously changing figure that your suggestions should give me.
> 
> hari
> 
> 
> 
> 
> Using plt.ion() or plt.ioff() causes a spinning blue-ball on windows..while the rest of the script continues.
> If I use the figure number trick. I get the first figure displayed.
> 
> On Tue, Oct 16, 2012 at 12:15 PM, Benjamin Root <ben...@ou...> wrote:
> 
> 
> On Tue, Oct 16, 2012 at 11:25 AM, hari jayaram <ha...@gm...> wrote:
> Hi 
> I am a relative newbie to matplotlib.
> 
> I have a python script that handles a dataset that comprises 384 sets of data.
> 
> At the present moment , I read in a set of data - process it - and the create a figure using code shown below.
> I am using windows with the default backend ( I think I set it to wx).
> 
> When I run the program, figure after figure shows up..the program continues from well to well plotting the figure. I can close the figure window using the X on the right -hand side..while the program chugs along.
> 
> Is there a way to just recycle the figure object , so that the plot shows up for a brief second and refreshes when the next calculation is complete. Each process_data function , takes a few minutes. 
> 
> Alternatively I just want to close the figure object I show after a brief lag. I am OK if that happens instantaneously..but I dont know how to achieve this.
> Do I have to use the matplotlib.Figure object to achieve this functionality
> 
> Thanks
> Hari
> 
> 
> 
> Hari,
> 
> To recycle the figure, try the following:
> 
> 
> 
> import matplotlib.pyplot as plt
> 
> def do_my_plot(par1, par2, well_id):
> processed_data_object = processed_dict[well_id]
> # Plot all the data
> par1.plot(processed_data_object.raw_x,processed_data_object.raw_y).
> par2.plot(....
> # finally
> plt.show() 
> # I tried fig.clf()
> 
> 
> def plot_and_process_data():
> plt.ion() # Turn on interactive mode 
> fig = plt.figure(figsize=(7,7)
> ax = fig.add_subplot(1,1,1)
> par1 =ax.twinx()
> par2 = ax.twinx()
> 
> for well_id in list_of_384_well_ids:
> par1.cla()
> par2.cla()
> process_data(well_id)
> do_my_plot(par1, par2, well_id)
> 
> Note, this is completely untested, but it would be how I would go about it at first. The "plt.ion()" turns on interactive mode to allow your code to continue running even after the plot window appears (but does not end until the last window is closed.). Of course, another approach would simply be to do "fig.savefig()" after every update to the figure and never use show() and ion() (essentially, a non-interactive head-less script).
> 
> Hopefully, this helps.
> Ben Root
> 
> 
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct_______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Hello group,
is there any special reason for the special handling of scalar 
arguments in the normalization classes? Would it not be simpler 
to just use asarray and only use arrays? 
If nothing speaks against it, i could do a pull request.
greetings
Till Stensitzki

On Wed, Oct 17, 2012 at 12:03 PM, Damon McDougall <dam...@gm...
> wrote:
>
>
> Also notice the triangle transparency...
>
>
True. Mike's 4 line addition fixes that issue:
https://github.com/matplotlib/matplotlib/issues/1410
-- 
Gökhan

I see that the same behavior here on 3 different viewers. It is a slight
aesthetic issue, but once in a while I come up similar differences between
PDF and PNGs outputs.
-- 
Gökhan

On Wed, Oct 17, 2012 at 6:58 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Wed, Oct 17, 2012 at 1:21 PM, Damon McDougall <dam...@gm...>
> wrote:
>>
>> On Wed, Oct 17, 2012 at 6:13 PM, Benjamin Root <ben...@ou...> wrote:
>> >
>> >
>> > On Wed, Oct 17, 2012 at 1:08 PM, Damon McDougall
>> > <dam...@gm...>
>> > wrote:
>> >>
>> >> On Wed, Oct 17, 2012 at 5:56 PM, Gökhan Sever <gok...@gm...>
>> >> wrote:
>> >> >
>> >> >
>> >> > On Wed, Oct 17, 2012 at 10:21 AM, Benjamin Root <ben...@ou...>
>> >> > wrote:
>> >> >>
>> >> >>
>> >> >>
>> >> >> On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever
>> >> >> <gok...@gm...>
>> >> >> wrote:
>> >> >>>
>> >> >>> Thanks Mike,
>> >> >>>
>> >> >>> Another point I noticed is setting linewidth to 0 (in fill_between
>> >> >>> function) isn't working as expected when figure is saved as a PDF
>> >> >>> file.
>> >> >>>
>> >> >>> I noticed this while posting a sample script on scipy-users:
>> >> >>>
>> >> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
>> >> >>>
>> >> >>> Compare the outputs of pdf and png to see the difference that I am
>> >> >>> mentioning:
>> >> >>>
>> >> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
>> >> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
>> >> >>>
>> >> >>
>> >> >> Actually, this is not a bug in mpl. It is a "bug" in various
>> >> >> viewers.
>> >> >> Some viewers have a "minimum" linewidth and will use that for any
>> >> >> requested
>> >> >> linewidths smaller than that. Are you using Apple's Preview?
>> >> >>
>> >> >> Cheers!
>> >> >> Ben Root
>> >> >>
>> >> >
>> >> > Hi Ben,
>> >> >
>> >> > I was guessing the same way, but trying different viewers doesn't
>> >> > make
>> >> > any
>> >> > difference:
>> >> >
>> >> > Evince v3.2.1
>> >> > Xpdf v3.03
>> >> > Okular v0.14.3
>> >> >
>> >> > Could anyone confirm this on a window machine?
>> >> >
>> >> > --
>> >> > Gökhan
>> >>
>> >> Confirmed with Preview.app on OS X 10.7.4.
>> >>
>> >
>> > I think it looks ok with the old Adobe viewer on linux (I never bothered
>> > updating it since I use Evince). Kinda hard to tell the difference
>> > between
>> > grey and greyer... Could somebody include a screenshot of what they are
>> > seeing?
>>
>> Sir, the internet has completed your request: http://i.imgur.com/UdRB9.png
>>
>> >
>> > Ben Root
>> >
>>
>>
>
> Thanks! That helps.
>
> With the older Adobe viewer for linux, there is a very slight line, but not
> nearly as pronounced as Apple's Preview (which is where I originally
> encountered this issue about a year ago). Attaching a screen capture for
> reference.
>
> Cheers!
> Ben Root
>
Also notice the triangle transparency...
-- 
Damon McDougall
http://www.damon-is-a-geek.com
B2.39
Mathematics Institute
University of Warwick
Coventry
West Midlands
CV4 7AL
United Kingdom

On Wed, Oct 17, 2012 at 1:21 PM, Damon McDougall
<dam...@gm...>wrote:
> On Wed, Oct 17, 2012 at 6:13 PM, Benjamin Root <ben...@ou...> wrote:
> >
> >
> > On Wed, Oct 17, 2012 at 1:08 PM, Damon McDougall <
> dam...@gm...>
> > wrote:
> >>
> >> On Wed, Oct 17, 2012 at 5:56 PM, Gökhan Sever <gok...@gm...>
> >> wrote:
> >> >
> >> >
> >> > On Wed, Oct 17, 2012 at 10:21 AM, Benjamin Root <ben...@ou...>
> wrote:
> >> >>
> >> >>
> >> >>
> >> >> On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever <
> gok...@gm...>
> >> >> wrote:
> >> >>>
> >> >>> Thanks Mike,
> >> >>>
> >> >>> Another point I noticed is setting linewidth to 0 (in fill_between
> >> >>> function) isn't working as expected when figure is saved as a PDF
> >> >>> file.
> >> >>>
> >> >>> I noticed this while posting a sample script on scipy-users:
> >> >>>
> >> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
> >> >>>
> >> >>> Compare the outputs of pdf and png to see the difference that I am
> >> >>> mentioning:
> >> >>>
> >> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
> >> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
> >> >>>
> >> >>
> >> >> Actually, this is not a bug in mpl. It is a "bug" in various
> viewers.
> >> >> Some viewers have a "minimum" linewidth and will use that for any
> >> >> requested
> >> >> linewidths smaller than that. Are you using Apple's Preview?
> >> >>
> >> >> Cheers!
> >> >> Ben Root
> >> >>
> >> >
> >> > Hi Ben,
> >> >
> >> > I was guessing the same way, but trying different viewers doesn't make
> >> > any
> >> > difference:
> >> >
> >> > Evince v3.2.1
> >> > Xpdf v3.03
> >> > Okular v0.14.3
> >> >
> >> > Could anyone confirm this on a window machine?
> >> >
> >> > --
> >> > Gökhan
> >>
> >> Confirmed with Preview.app on OS X 10.7.4.
> >>
> >
> > I think it looks ok with the old Adobe viewer on linux (I never bothered
> > updating it since I use Evince). Kinda hard to tell the difference
> between
> > grey and greyer... Could somebody include a screenshot of what they are
> > seeing?
>
> Sir, the internet has completed your request: http://i.imgur.com/UdRB9.png
>
> >
> > Ben Root
> >
>
>
>
Thanks! That helps.
With the older Adobe viewer for linux, there is a very slight line, but not
nearly as pronounced as Apple's Preview (which is where I originally
encountered this issue about a year ago). Attaching a screen capture for
reference.
Cheers!
Ben Root

Thanks Benjamin, Sterling and Damon for your prompt help
However I am still not able to achieve what I wanted .
I can get the headless script to work just great where it saves all the
figures and I can view them after the script is done running.
But somehow when I try the figure number method that Sterling suggested ,
along with the axis clear and redraw method (Damon) , or the decouple and
clear and then plot method (Benjamin Root) : I get the plot just spinning
with a blue circle on Windows 7 and the script just chugs merrily along.
I think part of the problem was that I was wrong in the way I stated my
application. Each of the 384 data processing steps takes a few seconds..and
not a minute as I had indicated. I tried with both ion() and ioff() and
giving the figure a number , which stays constant and clearing the axis
everytime before plotting. But I get a spiining blue circle in Windows.
I will try and cookup a test case , and send to the list , to reproduce
what I am seeing. it may still be that I am calling pylab , pyplot
incorrectly and hence not getting the continuously changing figure that
your suggestions should give me.
hari
Using plt.ion() or plt.ioff() causes a spinning blue-ball on windows..while
the rest of the script continues.
If I use the figure number trick. I get the first figure displayed.
On Tue, Oct 16, 2012 at 12:15 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Tue, Oct 16, 2012 at 11:25 AM, hari jayaram <ha...@gm...> wrote:
>
>> Hi
>> I am a relative newbie to matplotlib.
>>
>> I have a python script that handles a dataset that comprises 384 sets of
>> data.
>>
>> At the present moment , I read in a set of data - process it - and the
>> create a figure using code shown below.
>> I am using windows with the default backend ( I think I set it to wx).
>>
>> When I run the program, figure after figure shows up..the program
>> continues from well to well plotting the figure. I can close the figure
>> window using the X on the right -hand side..while the program chugs along.
>>
>> Is there a way to just recycle the figure object , so that the plot shows
>> up for a brief second and refreshes when the next calculation is complete.
>> Each process_data function , takes a few minutes.
>>
>> Alternatively I just want to close the figure object I show after a brief
>> lag. I am OK if that happens instantaneously..but I dont know how
>> to achieve this.
>> Do I have to use the matplotlib.Figure object to achieve this
>> functionality
>>
>> Thanks
>> Hari
>>
>>
>>
> Hari,
>
> To recycle the figure, try the following:
>
>
>
> import matplotlib.pyplot as plt
>
> def do_my_plot(par1, par2, well_id):
> processed_data_object = processed_dict[well_id]
> # Plot all the data
> par1.plot(processed_data_object.raw_x,processed_data_object.raw_y).
> par2.plot(....
> # finally
> plt.show()
> # I tried fig.clf()
>
>
> def plot_and_process_data():
> plt.ion() # Turn on interactive mode
> fig = plt.figure(figsize=(7,7)
> ax = fig.add_subplot(1,1,1)
> par1 =ax.twinx()
> par2 = ax.twinx()
>
> for well_id in list_of_384_well_ids:
> par1.cla()
> par2.cla()
> process_data(well_id)
> do_my_plot(par1, par2, well_id)
>
> Note, this is completely untested, but it would be how I would go about it
> at first. The "plt.ion()" turns on interactive mode to allow your code to
> continue running even after the plot window appears (but does not end until
> the last window is closed.). Of course, another approach would simply be
> to do "fig.savefig()" after every update to the figure and never use show()
> and ion() (essentially, a non-interactive head-less script).
>
> Hopefully, this helps.
> Ben Root
>
>

On Wed, Oct 17, 2012 at 6:13 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Wed, Oct 17, 2012 at 1:08 PM, Damon McDougall <dam...@gm...>
> wrote:
>>
>> On Wed, Oct 17, 2012 at 5:56 PM, Gökhan Sever <gok...@gm...>
>> wrote:
>> >
>> >
>> > On Wed, Oct 17, 2012 at 10:21 AM, Benjamin Root <ben...@ou...> wrote:
>> >>
>> >>
>> >>
>> >> On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever <gok...@gm...>
>> >> wrote:
>> >>>
>> >>> Thanks Mike,
>> >>>
>> >>> Another point I noticed is setting linewidth to 0 (in fill_between
>> >>> function) isn't working as expected when figure is saved as a PDF
>> >>> file.
>> >>>
>> >>> I noticed this while posting a sample script on scipy-users:
>> >>>
>> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
>> >>>
>> >>> Compare the outputs of pdf and png to see the difference that I am
>> >>> mentioning:
>> >>>
>> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
>> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
>> >>>
>> >>
>> >> Actually, this is not a bug in mpl. It is a "bug" in various viewers.
>> >> Some viewers have a "minimum" linewidth and will use that for any
>> >> requested
>> >> linewidths smaller than that. Are you using Apple's Preview?
>> >>
>> >> Cheers!
>> >> Ben Root
>> >>
>> >
>> > Hi Ben,
>> >
>> > I was guessing the same way, but trying different viewers doesn't make
>> > any
>> > difference:
>> >
>> > Evince v3.2.1
>> > Xpdf v3.03
>> > Okular v0.14.3
>> >
>> > Could anyone confirm this on a window machine?
>> >
>> > --
>> > Gökhan
>>
>> Confirmed with Preview.app on OS X 10.7.4.
>>
>
> I think it looks ok with the old Adobe viewer on linux (I never bothered
> updating it since I use Evince). Kinda hard to tell the difference between
> grey and greyer... Could somebody include a screenshot of what they are
> seeing?
Sir, the internet has completed your request: http://i.imgur.com/UdRB9.png
>
> Ben Root
>
-- 
Damon McDougall
http://www.damon-is-a-geek.com
B2.39
Mathematics Institute
University of Warwick
Coventry
West Midlands
CV4 7AL
United Kingdom

On Wed, Oct 17, 2012 at 1:08 PM, Damon McDougall
<dam...@gm...>wrote:
> On Wed, Oct 17, 2012 at 5:56 PM, Gökhan Sever <gok...@gm...>
> wrote:
> >
> >
> > On Wed, Oct 17, 2012 at 10:21 AM, Benjamin Root <ben...@ou...> wrote:
> >>
> >>
> >>
> >> On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever <gok...@gm...>
> >> wrote:
> >>>
> >>> Thanks Mike,
> >>>
> >>> Another point I noticed is setting linewidth to 0 (in fill_between
> >>> function) isn't working as expected when figure is saved as a PDF file.
> >>>
> >>> I noticed this while posting a sample script on scipy-users:
> >>>
> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
> >>>
> >>> Compare the outputs of pdf and png to see the difference that I am
> >>> mentioning:
> >>>
> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
> >>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
> >>>
> >>
> >> Actually, this is not a bug in mpl. It is a "bug" in various viewers.
> >> Some viewers have a "minimum" linewidth and will use that for any
> requested
> >> linewidths smaller than that. Are you using Apple's Preview?
> >>
> >> Cheers!
> >> Ben Root
> >>
> >
> > Hi Ben,
> >
> > I was guessing the same way, but trying different viewers doesn't make
> any
> > difference:
> >
> > Evince v3.2.1
> > Xpdf v3.03
> > Okular v0.14.3
> >
> > Could anyone confirm this on a window machine?
> >
> > --
> > Gökhan
>
> Confirmed with Preview.app on OS X 10.7.4.
>
>
I think it looks ok with the old Adobe viewer on linux (I never bothered
updating it since I use Evince). Kinda hard to tell the difference between
grey and greyer... Could somebody include a screenshot of what they are
seeing?
Ben Root

On Wed, Oct 17, 2012 at 5:56 PM, Gökhan Sever <gok...@gm...> wrote:
>
>
> On Wed, Oct 17, 2012 at 10:21 AM, Benjamin Root <ben...@ou...> wrote:
>>
>>
>>
>> On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever <gok...@gm...>
>> wrote:
>>>
>>> Thanks Mike,
>>>
>>> Another point I noticed is setting linewidth to 0 (in fill_between
>>> function) isn't working as expected when figure is saved as a PDF file.
>>>
>>> I noticed this while posting a sample script on scipy-users:
>>>
>>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
>>>
>>> Compare the outputs of pdf and png to see the difference that I am
>>> mentioning:
>>>
>>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
>>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
>>>
>>
>> Actually, this is not a bug in mpl. It is a "bug" in various viewers.
>> Some viewers have a "minimum" linewidth and will use that for any requested
>> linewidths smaller than that. Are you using Apple's Preview?
>>
>> Cheers!
>> Ben Root
>>
>
> Hi Ben,
>
> I was guessing the same way, but trying different viewers doesn't make any
> difference:
>
> Evince v3.2.1
> Xpdf v3.03
> Okular v0.14.3
>
> Could anyone confirm this on a window machine?
>
> --
> Gökhan
Confirmed with Preview.app on OS X 10.7.4.
-- 
Damon McDougall
http://www.damon-is-a-geek.com
B2.39
Mathematics Institute
University of Warwick
Coventry
West Midlands
CV4 7AL
United Kingdom

On Wed, Oct 17, 2012 at 10:21 AM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever <gok...@gm...>wrote:
>
>> Thanks Mike,
>>
>> Another point I noticed is setting linewidth to 0 (in fill_between
>> function) isn't working as expected when figure is saved as a PDF file.
>>
>> I noticed this while posting a sample script on scipy-users:
>>
>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
>>
>> Compare the outputs of pdf and png to see the difference that I am
>> mentioning:
>>
>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
>> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
>>
>>
> Actually, this is not a bug in mpl. It is a "bug" in various viewers.
> Some viewers have a "minimum" linewidth and will use that for any requested
> linewidths smaller than that. Are you using Apple's Preview?
>
> Cheers!
> Ben Root
>
>
Hi Ben,
I was guessing the same way, but trying different viewers doesn't make any
difference:
Evince v3.2.1
Xpdf v3.03
Okular v0.14.3
Could anyone confirm this on a window machine?
-- 
Gökhan

On 10/16/12 12:29 PM, Jeff Whitaker wrote:
> On 10/16/12 11:20 AM, Rich Signell wrote:
>> Jeff,
>> Yep, that worked! So here is a working example of OWSlib with
>> Basemap: http://nbviewer.ipython.org/3900648/
>>
>> I switched the Basemap projection to 'cyl' because we need to ensure
>> that Basemap and WMS are using the same projection, right? (and since
>> I had requested EPSG:4326 from WMS, that's the 'cyl' in Basemap).
>>
>> Thanks!
>> Rich
> Rich: I took your code and made it into a new example.
>
> https://github.com/matplotlib/basemap/pull/84
>
> I think it may be better not to try to create a wmsimage method, since
> OWSlib.wms.WebMapService is quite a complicated beast to wrap.
>
> -Jeff
Rich: I went ahead and added a wmsimage method to Basemap (similar to 
Klo's implementation) and modified your example to use it. The extra 
**kwargs are just passed on to OWSLib.wms.WebMapService.getmap. Please 
add comments to pull request
https://github.com/matplotlib/basemap/pull/84
-Jeff
>> On Tue, Oct 16, 2012 at 11:38 AM, Jeff Whitaker <js...@fa...> wrote:
>>> On 10/16/12 8:48 AM, Rich Signell wrote:
>>>> Klo & Jeff,
>>>>
>>>> I tried making a concrete example of using OWSlib with Basemap, but
>>>> althought the WMS image looks good, the warpimage does not.
>>>>
>>>> http://nbviewer.ipython.org/3899690/
>>>>
>>>> Do you see where I went wrong?
>>>>
>>>> Thanks,
>>>> Rich
>>> Rich: warpimage assumes the image is of global extent. In your example, I
>>> think you can just pass the image to the basemap imshow method with
>>>
>>> from matplotlib.image import imread
>>> import urllib2
>>> m.imshow(imread(urllib2.urlopen(url)),origin='upper')
>>>
>>> Klo previously mentioned there might be a problem with the png data from the
>>> WMS server being 'chunked', s you might have to use klo's imshow_chunked
>>> function
>>>
>>> http://www.mail-archive.com/mat...@li.../msg25618.html
>>>
>>>
>>> -Jeff
>>>
>>>> On Sat, Oct 13, 2012 at 7:26 PM, klo uo <kl...@gm...> wrote:
>>>>> That's also what that snippet I linked does. You can add it to to Basemap
>>>>> and it should work.
>>>>>
>>>>> However Jeff suggested we use this tiny package OWSlib and handle WMS
>>>>> that
>>>>> way, which is better IMHO, but for some reason we did not got further
>>>>> reply.
>>>>>
>>>>>
>>>>>
>>>>> On Fri, Oct 12, 2012 at 1:31 PM, Rich Signell <rsi...@us...> wrote:
>>>>>> WMS services are required to respond to "GetCapabiltiies" request,
>>>>>> reporting what layers, styles, times, elevations, and projections they
>>>>>> have available. So for example, using the Unidata WMS example below,
>>>>>> if we do:
>>>>>>
>>>>>>
>>>>>>
>>>>>> http://motherlode.ucar.edu:8080/thredds/wms/fmrc/NCEP/NAM/CONUS_12km/NCEP-NAM-CONUS_12km-noaaport_best.ncd?service=WMS&request=GetCapabilities
>>>>>>
>>>>>> we can see from the XML response that the Coordinate Reference Systems
>>>>>> supported are:
>>>>>>
>>>>>> <CRS>EPSG:4326</CRS>
>>>>>> <CRS>CRS:84</CRS>
>>>>>> <CRS>EPSG:41001</CRS>
>>>>>> <CRS>EPSG:3857</CRS>
>>>>>> <CRS>EPSG:27700</CRS>
>>>>>> <CRS>EPSG:3408</CRS>
>>>>>> <CRS>EPSG:3409</CRS>
>>>>>> <CRS>EPSG:32661</CRS>
>>>>>> <CRS>EPSG:32761</CRS>
>>>>>>
>>>>>> And for this server, the supported response types are:
>>>>>> <Format>image/jpeg</Format>
>>>>>> <Format>image/png</Format>
>>>>>> <Format>application/vnd.google-earth.kmz</Format>
>>>>>> <Format>image/gif</Format>
>>>>>>
>>>>>> So I guess one way to proceed if you wanted to use WMS in Matplotlib
>>>>>> and avoid reprojection in python would be to:
>>>>>> 1. do the WMS GetCapabilities request to find the available supported
>>>>>> Coordinate Reference Systems (which will vary with WMS server)
>>>>>> 2. setup Basemap to use one of these CRS
>>>>>> 3. use the bounding box of your current axis (in projection units) as
>>>>>> part of a GetMap request to the WMS.
>>>>>>
>>>>>> -Rich
>>>>>>
>>>>>> On Thu, Oct 11, 2012 at 12:16 AM, klo uo <kl...@gm...> wrote:
>>>>>>> I guess that's it?
>>>>>>>
>>>>>>> warpimage() as it is now, checks if passed image is url, so we can add
>>>>>>> additional check if image is url, with urlparse to deduce image
>>>>>>> coordinates
>>>>>>> and projection if present, then overlay it over already created Basemap
>>>>>>> object.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> ------------------------------------------------------------------------------
>>>>>>> Don't let slow site performance ruin your business. Deploy New Relic
>>>>>>> APM
>>>>>>> Deploy New Relic app performance management and know exactly
>>>>>>> what is happening inside your Ruby, Python, PHP, Java, and .NET app
>>>>>>> Try New Relic at no cost today and get our sweet Data Nerd shirt too!
>>>>>>> http://p.sf.net/sfu/newrelic-dev2dev
>>>>>>> _______________________________________________
>>>>>>> Matplotlib-users mailing list
>>>>>>> Mat...@li...
>>>>>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>>>>>>
>>>>>> --
>>>>>> Dr. Richard P. Signell (508) 457-2229
>>>>>> USGS, 384 Woods Hole Rd.
>>>>>> Woods Hole, MA 02543-1598
>>
>
> ------------------------------------------------------------------------------
> Don't let slow site performance ruin your business. Deploy New Relic APM
> Deploy New Relic app performance management and know exactly
> what is happening inside your Ruby, Python, PHP, Java, and .NET app
> Try New Relic at no cost today and get our sweet Data Nerd shirt too!
> http://p.sf.net/sfu/newrelic-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
-- 
Jeffrey S. Whitaker Phone : (303)497-6313
Meteorologist FAX : (303)497-6449
NOAA/OAR/PSD R/PSD1 Email : Jef...@no...
325 Broadway Office : Skaggs Research Cntr 1D-113
Boulder, CO, USA 80303-3328 Web : http://tinyurl.com/5telg

On Wed, Oct 17, 2012 at 12:17 PM, Gökhan Sever <gok...@gm...>wrote:
> Thanks Mike,
>
> Another point I noticed is setting linewidth to 0 (in fill_between
> function) isn't working as expected when figure is saved as a PDF file.
>
> I noticed this while posting a sample script on scipy-users:
>
> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
>
> Compare the outputs of pdf and png to see the difference that I am
> mentioning:
>
> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
> http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
>
>
Actually, this is not a bug in mpl. It is a "bug" in various viewers.
Some viewers have a "minimum" linewidth and will use that for any requested
linewidths smaller than that. Are you using Apple's Preview?
Cheers!
Ben Root

Thanks Mike,
Another point I noticed is setting linewidth to 0 (in fill_between
function) isn't working as expected when figure is saved as a PDF file.
I noticed this while posting a sample script on scipy-users:
http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.py
Compare the outputs of pdf and png to see the difference that I am
mentioning:
http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.pdf
http://atmos.uwyo.edu/~gsever/data/test/curvefit_test.png
On Wed, Oct 17, 2012 at 9:16 AM, Michael Droettboom <md...@st...> wrote:
> Filed as https://github.com/matplotlib/matplotlib/issues/1410
>
>
> On 10/16/2012 10:38 PM, Eric Firing wrote:
>
> On 2012年10月16日 4:27 PM, Gökhan Sever wrote:
>
> Hello,
>
> I see that a few days old clone of mpl, cannot save open symbols
> correctly in a pdf file.
>
> Here is a simple test case (in ipython --pylab):
>
> I6 xx = np.random.random(1000)
>
> I7 plt.plot(xx, 'D', mfc='none')
>
> On screen open symbols are fine, as expected transparency works fine,
> however when saved in pdf transparency disappears, symbols cover each
> other. (Saving to png is also fine)
>
> Could you confirm if this is a case in another installation?
>
> Confirmed with 1.2.x. The problem is only on pdf; ps and svg are correct.
>
> Eric
>
>
> Thanks.
>
> --
> Gökhan
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:http://p.sf.net/sfu/appdyn_sfd2d_oct
>
>
>
> _______________________________________________
> Matplotlib-users mailing lis...@li...https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-users mailing lis...@li...https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
-- 
Gökhan

I also think that would be useful. It would, for example, allow to
generate "preview" plots from other languages, without interfacing
them to Python. It must be said that MPL is actually quite nice
looking in the default settings for basic plotting, and this is an
nice feature that can exploit.
I have seen many pieces of code using SuperMongo for plotting. I find
it absolutely ugly, and super expensive, but it is probably the easier
plotting system to call from FORTRAN (or, at least, it was).
On Wed, Oct 17, 2012 at 3:15 PM, Michael Droettboom <md...@st...> wrote:
> I think this could be very useful -- and might help increase the user
> base beyond the Python community. As I, and most of us on this list,
> are quite comfortable with Python, I don't think I'd use it myself, but
> I certainly see the utility of it.
>
> Mike
>
> On 10/17/2012 06:38 AM, Damon McDougall wrote:
>> All,
>>
>> I was brain-storming yesterday and I wanted to test the waters to see
>> if people would find it useful.
>>
>> Currently, GNU plotutils comes with command-line utilities such as
>> `graph` to create quick and dirty line plots like this:
>> http://www.gnu.org/software/gsl/manual/html_node/DWT-Examples.html. I
>> think even gnuplot might be similar.
>>
>> How do people feel about perhaps adding a matplotlib version, mocking
>> the same calling signature as graph?
>>
>> I think the most important question is: would it be useful?
>>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Filed as https://github.com/matplotlib/matplotlib/issues/1410
On 10/16/2012 10:38 PM, Eric Firing wrote:
> On 2012年10月16日 4:27 PM, Gökhan Sever wrote:
>> Hello,
>>
>> I see that a few days old clone of mpl, cannot save open symbols
>> correctly in a pdf file.
>>
>> Here is a simple test case (in ipython --pylab):
>>
>> I6 xx = np.random.random(1000)
>>
>> I7 plt.plot(xx, 'D', mfc='none')
>>
>> On screen open symbols are fine, as expected transparency works fine,
>> however when saved in pdf transparency disappears, symbols cover each
>> other. (Saving to png is also fine)
>>
>> Could you confirm if this is a case in another installation?
> Confirmed with 1.2.x. The problem is only on pdf; ps and svg are correct.
>
> Eric
>
>> Thanks.
>>
>> --
>> Gökhan
>>
>>
>> ------------------------------------------------------------------------------
>> Everyone hates slow websites. So do we.
>> Make your web apps faster with AppDynamics
>> Download AppDynamics Lite for free today:
>> http://p.sf.net/sfu/appdyn_sfd2d_oct
>>
>>
>>
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

I think this could be very useful -- and might help increase the user 
base beyond the Python community. As I, and most of us on this list, 
are quite comfortable with Python, I don't think I'd use it myself, but 
I certainly see the utility of it.
Mike
On 10/17/2012 06:38 AM, Damon McDougall wrote:
> All,
>
> I was brain-storming yesterday and I wanted to test the waters to see
> if people would find it useful.
>
> Currently, GNU plotutils comes with command-line utilities such as
> `graph` to create quick and dirty line plots like this:
> http://www.gnu.org/software/gsl/manual/html_node/DWT-Examples.html. I
> think even gnuplot might be similar.
>
> How do people feel about perhaps adding a matplotlib version, mocking
> the same calling signature as graph?
>
> I think the most important question is: would it be useful?
>

On Wed, Oct 17, 2012 at 1:41 AM, Eric Firing <ef...@ha...> wrote:
> sudo apt-get build-dep python-matplotlib
>
> It might pull in more than you really want, but it will certainly
> include gui toolkits.
If you want to avoid installing all the packages that you get with the
`apt-get build-dep pythoh-matplotlib`, the following seems to be
enough:
sudo apt-get install libfreetype6-dev libpng12-dev python-tk tk-dev
python-gtk2-dev
Alejandro.

All,
I was brain-storming yesterday and I wanted to test the waters to see
if people would find it useful.
Currently, GNU plotutils comes with command-line utilities such as
`graph` to create quick and dirty line plots like this:
http://www.gnu.org/software/gsl/manual/html_node/DWT-Examples.html. I
think even gnuplot might be similar.
How do people feel about perhaps adding a matplotlib version, mocking
the same calling signature as graph?
I think the most important question is: would it be useful?
-- 
Damon McDougall
http://www.damon-is-a-geek.com
B2.39
Mathematics Institute
University of Warwick
Coventry
West Midlands
CV4 7AL
United Kingdom

2012年10月17日 Damon McDougall <dam...@gm...>
> On Wed, Oct 17, 2012 at 9:12 AM, Francesco Montesano
> <fra...@gm...> wrote:
> >
> >
> > 2012年10月17日 Eric Firing <ef...@ha...>
> >>
> >> On 2012年10月16日 9:22 PM, Francesco Montesano wrote:
> >> > Dear list,
> >> >
> >> > I've see a difference between the default backend between
> >> >
> >> > v1.1.1 (shipped with kubuntu 12.10dev) and v1.2.0.rc1, 1.2.0rc2 and
> >> > master (1.3.x).
> >> >
> >> >
> >> > My set up is to call ipython with pylab and turn on interactive mode.
> I
> >> > still haven't copied over my matplotlibrc file from my work computer
> >> > (there I use qtagg, if I remember well)
> >> >
> >> >
> >> > On v1.1.1 the default is [backend: TkAgg], while in the other two
> cases
> >> > it is [backend: agg]. Is there some reason for this difference?
> >>
> >> The default should be based on what is found when mpl is built, so it
> >> sounds like when you are building v1.2.x, none of the supported gui
> >> toolkits is being found. I don't know why that is. Are you installing
> >> on a system that has gui toolkits already installed?
> >
> > I think so, as the v1.1.1 shipped with my OS is using TkAgg as backend
> and I
> > can plot interactively.
> >
> > When I build from source I get
> > OPTIONAL BACKEND DEPENDENCIES
> > libpng: found, but unknown version (no pkg-config)
> > Tkinter: no
> > * Using default library and include directories
> for
> > * Tcl and Tk because a Tk window failed to open.
> > * You may need to define DISPLAY for Tk to work
> so
> > * that setup can determine where your libraries
> are
> > * located. Tkinter present, but header files are
> not
> > * found. You may need to install development
> > * packages.
> > * You may need to install 'dev' package(s) to
> > * provide header files.
> > Gtk+: no
> > * Could not find Gtk+ headers in any of
> > * '/usr/local/include', '/usr/include',
> > * '/usr/local/include', '/usr/include', '.'
> > Mac OS X native: no
> > Qt: no
> > Qt4: Qt: 4.8.2, PyQt4: 4.9.3
> > PySide: no
> > Cairo: 1.8.8
> >
> > Could be a problem with Tkinter and dev packages.
> >
> >
> >>
> >>
> >> One useful
> >> technique with ubuntu derivatives is to do
> >>
> >> sudo apt-get build-dep python-matplotlib
> >>
> > probably it's easier if I just switch to qtagg, as I already have it
> >
> >>
> >> It might pull in more than you really want, but it will certainly
> >> include gui toolkits.
> >>
> >> Eric
> >
> > Thanks,
> > Francesco
> >
> >>
> >>
> >> >
> >> >
> >> > cheers,
> >> >
> >> > Francesco
>
> Packages have whatever default backend the maintainer decided to build
> them with. For example, the macports packaged version of matplotlib is
> maintained by someone who specifies the macosx backend to be the
> default. When building from source, I think the TkAgg backend is the
> default, because most platforms have tkinter installed out of the box.
> Since it appears you don't have Tkinter installed, Agg is the default.
>
> Agg is a non-gui backend (but it produces awesome output).
>
Thanks for the explanation.
Fra
>
> --
> Damon McDougall
> http://www.damon-is-a-geek.com
> B2.39
> Mathematics Institute
> University of Warwick
> Coventry
> West Midlands
> CV4 7AL
> United Kingdom
>

On Wed, Oct 17, 2012 at 9:12 AM, Francesco Montesano
<fra...@gm...> wrote:
>
>
> 2012年10月17日 Eric Firing <ef...@ha...>
>>
>> On 2012年10月16日 9:22 PM, Francesco Montesano wrote:
>> > Dear list,
>> >
>> > I've see a difference between the default backend between
>> >
>> > v1.1.1 (shipped with kubuntu 12.10dev) and v1.2.0.rc1, 1.2.0rc2 and
>> > master (1.3.x).
>> >
>> >
>> > My set up is to call ipython with pylab and turn on interactive mode. I
>> > still haven't copied over my matplotlibrc file from my work computer
>> > (there I use qtagg, if I remember well)
>> >
>> >
>> > On v1.1.1 the default is [backend: TkAgg], while in the other two cases
>> > it is [backend: agg]. Is there some reason for this difference?
>>
>> The default should be based on what is found when mpl is built, so it
>> sounds like when you are building v1.2.x, none of the supported gui
>> toolkits is being found. I don't know why that is. Are you installing
>> on a system that has gui toolkits already installed?
>
> I think so, as the v1.1.1 shipped with my OS is using TkAgg as backend and I
> can plot interactively.
>
> When I build from source I get
> OPTIONAL BACKEND DEPENDENCIES
> libpng: found, but unknown version (no pkg-config)
> Tkinter: no
> * Using default library and include directories for
> * Tcl and Tk because a Tk window failed to open.
> * You may need to define DISPLAY for Tk to work so
> * that setup can determine where your libraries are
> * located. Tkinter present, but header files are not
> * found. You may need to install development
> * packages.
> * You may need to install 'dev' package(s) to
> * provide header files.
> Gtk+: no
> * Could not find Gtk+ headers in any of
> * '/usr/local/include', '/usr/include',
> * '/usr/local/include', '/usr/include', '.'
> Mac OS X native: no
> Qt: no
> Qt4: Qt: 4.8.2, PyQt4: 4.9.3
> PySide: no
> Cairo: 1.8.8
>
> Could be a problem with Tkinter and dev packages.
>
>
>>
>>
>> One useful
>> technique with ubuntu derivatives is to do
>>
>> sudo apt-get build-dep python-matplotlib
>>
> probably it's easier if I just switch to qtagg, as I already have it
>
>>
>> It might pull in more than you really want, but it will certainly
>> include gui toolkits.
>>
>> Eric
>
> Thanks,
> Francesco
>
>>
>>
>> >
>> >
>> > cheers,
>> >
>> > Francesco
Packages have whatever default backend the maintainer decided to build
them with. For example, the macports packaged version of matplotlib is
maintained by someone who specifies the macosx backend to be the
default. When building from source, I think the TkAgg backend is the
default, because most platforms have tkinter installed out of the box.
Since it appears you don't have Tkinter installed, Agg is the default.
Agg is a non-gui backend (but it produces awesome output).
-- 
Damon McDougall
http://www.damon-is-a-geek.com
B2.39
Mathematics Institute
University of Warwick
Coventry
West Midlands
CV4 7AL
United Kingdom

2012年10月17日 Eric Firing <ef...@ha...>
> On 2012年10月16日 9:22 PM, Francesco Montesano wrote:
> > Dear list,
> >
> > I've see a difference between the default backend between
> >
> > v1.1.1 (shipped with kubuntu 12.10dev) and v1.2.0.rc1, 1.2.0rc2 and
> > master (1.3.x).
> >
> >
> > My set up is to call ipython with pylab and turn on interactive mode. I
> > still haven't copied over my matplotlibrc file from my work computer
> > (there I use qtagg, if I remember well)
> >
> >
> > On v1.1.1 the default is [backend: TkAgg], while in the other two cases
> > it is [backend: agg]. Is there some reason for this difference?
>
> The default should be based on what is found when mpl is built, so it
> sounds like when you are building v1.2.x, none of the supported gui
> toolkits is being found. I don't know why that is. Are you installing
> on a system that has gui toolkits already installed?
I think so, as the v1.1.1 shipped with my OS is using TkAgg as backend and
I can plot interactively.
When I build from source I get
OPTIONAL BACKEND DEPENDENCIES
 libpng: found, but unknown version (no pkg-config)
 Tkinter: no
 * Using default library and include directories for
 * Tcl and Tk because a Tk window failed to open.
 * You may need to define DISPLAY for Tk to work so
 * that setup can determine where your libraries are
 * located. Tkinter present, but header files are not
 * found. You may need to install development
 * packages.
 * You may need to install 'dev' package(s) to
 * provide header files.
 Gtk+: no
 * Could not find Gtk+ headers in any of
 * '/usr/local/include', '/usr/include',
 * '/usr/local/include', '/usr/include', '.'
 Mac OS X native: no
 Qt: no
 Qt4: Qt: 4.8.2, PyQt4: 4.9.3
 PySide: no
 Cairo: 1.8.8
Could be a problem with Tkinter and dev packages.
>
> One useful
> technique with ubuntu derivatives is to do
sudo apt-get build-dep python-matplotlib
>
> probably it's easier if I just switch to qtagg, as I already have it
> It might pull in more than you really want, but it will certainly
> include gui toolkits.
>
> Eric
>
Thanks,
Francesco
>
> >
> >
> > cheers,
> >
> > Francesco
> >
> >
> >
> >
> ------------------------------------------------------------------------------
> > Everyone hates slow websites. So do we.
> > Make your web apps faster with AppDynamics
> > Download AppDynamics Lite for free today:
> > http://p.sf.net/sfu/appdyn_sfd2d_oct
> >
> >
> >
> > _______________________________________________
> > Matplotlib-users mailing list
> > Mat...@li...
> > https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> >
>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

On 16 October 2012 18:44, T J <tj...@gm...> wrote:
>
> This is a set of 152 points on a triangle. delaunay is mentioned to
> have problems for some pathological cases. Is a complete triangular
> grid considered as such a case?
>
Yes, under certain circumstances! delaunay is not 'geometrically robust',
meaning it doesn't take into account machine precision errors when deciding
if a point is on one side of a line or not or inside a circle or not, which
can result in errors leading to the algorithm failing. It sounds like this
should be an easy problem to solve but it isn't.
In your example the y-values that presumably should be in a straight line
about y=-0.5 feature variations at about the 16th decimal place that are
causing the algorithm to fail. If these variations were much larger or
much smaller, the algorithm would probably work. The first time the
failure occurs is at the 20th point, which you can see if you add a call to
something like
 plt.tripcolor(x, y, np.zeros_like(x), alpha=0.5)
The darker blue colour shows two triangles overlapping, which shouldn't
happen.
You have a number of options:
1) Round your x and y values to a smaller number of decimal places, e.g.
 x = x.round(8)
 y = y.round(8)
On my machine 8 decimal places is good, 10 isn't.
2) Add extra noise to your point positions, which is the classic way to get
around this delaunay limitation, e.g.
 x = x + 1e-6*np.random.rand(len(x))
 y = y + 1e-6*np.random.rand(len(x))
3) Specify the triangulation yourself, which is usually pretty easy for
regular cases like yours. See tricontour_demo.py for an example of how to
specify a triangulation in your tri* function calls.
I hope this helps,
Ian

On 2012年10月16日 9:22 PM, Francesco Montesano wrote:
> Dear list,
>
> I've see a difference between the default backend between
>
> v1.1.1 (shipped with kubuntu 12.10dev) and v1.2.0.rc1, 1.2.0rc2 and
> master (1.3.x).
>
>
> My set up is to call ipython with pylab and turn on interactive mode. I
> still haven't copied over my matplotlibrc file from my work computer
> (there I use qtagg, if I remember well)
>
>
> On v1.1.1 the default is [backend: TkAgg], while in the other two cases
> it is [backend: agg]. Is there some reason for this difference?
The default should be based on what is found when mpl is built, so it 
sounds like when you are building v1.2.x, none of the supported gui 
toolkits is being found. I don't know why that is. Are you installing 
on a system that has gui toolkits already installed? One useful 
technique with ubuntu derivatives is to do
sudo apt-get build-dep python-matplotlib
It might pull in more than you really want, but it will certainly 
include gui toolkits.
Eric
>
>
> cheers,
>
> Francesco
>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_sfd2d_oct
>
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>