matplotlib-devel Mailing List for matplotlib

Here it is.

Did it slightly different by calling the option "bars_on_top=False", which is 
more descriptive.

Be aware, that this changes the default behaviour, but the difference is so 
subtle that only few people should notice at all. Still, I think this is the 
more reasonable default behaviour, since - usually - the plot line is more 
important than the errorbars and should be clearly visible.



On Friday 22 October 2004 15:24, John Hunter wrote:
> How about adding a kwarg to the signature of the errorbar function,
> which defaults to below=True (this would have the default behavior you
> want, errorbars below markers, but could be customized).
>
> Then you could add an if clause in the errorbar code which reverses
> the order of the calls.
>
> Since noone has done the zordering yet, and this issue has come up
> twice in the context of errorbars, it is probably worthwhile to add it
> now to the errorbar code.
>
> If I could impose in you one more time Norbert to submit a patch that
> allows the user to configure the order using a kwarg, I'll be happy to
> check it in.

-- 
_________________________________________Norbert Nemec
         Bernhardstr. 2 ... D-93053 Regensburg
     Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
           eMail: <No...@Ne...>

>>>>> "Norbert" == Norbert Nemec <Nor...@gm...> writes:

    Norbert> Thanks for the reply. Surprises me that people would
    Norbert> really argue about this issue, but well. :-)

There's just no accounting for the taste of some people...  Here are
some links to the original discussion

  http://sourceforge.net/mailarchive/forum.php?thread_id=5434727&forum_id=33405
  http://sourceforge.net/mailarchive/forum.php?thread_id=5440722&forum_id=33405

    Norbert> Attached is a slighly changed patch: effectively, it does
    Norbert> not change the layering any more, but it changes the
    Norbert> internals of errorbar() in a way, that allows to swap the
    Norbert> layering by simply swapping two blocks of code. (Which
    Norbert> was nontrivial before, since the color from the plot line
    Norbert> can only be read after it has been drawn.)

    Norbert> Whether this swapping is then done by hand, by an option
    Norbert> or an entry in the rc file can be left to the future.

    Norbert> Maybe, submitting this patch will save someone else the
    Norbert> time to figure it out himself.

How about adding a kwarg to the signature of the errorbar function,
which defaults to below=True (this would have the default behavior you
want, errorbars below markers, but could be customized).

Then you could add an if clause in the errorbar code which reverses
the order of the calls.

Since noone has done the zordering yet, and this issue has come up
twice in the context of errorbars, it is probably worthwhile to add it
now to the errorbar code.

If I could impose in you one more time Norbert to submit a patch that
allows the user to configure the order using a kwarg, I'll be happy to
check it in.

BTW, Gary, any chance I could persuade you to write a tutorial/guide
to using errorbar / bar / barh for the users guide?  barh is in CVS.
If you are averse to latex, you could submit in plain text and I can
convert it.  I'm mostly done with the matlab interface section and
have been working on other sections (API) and I am having trouble
motivating myself to finish some of these matlab interface sections.
Since you are the world's expert on matplotlib bar charts, I thought
you would be a good victim, er candidate, to write that section.  FYI,
I looked into changing the default meaning of x in the errorbar code
to have it refer to the center, rather than the left, as we discussed.
The problem is, this broke some of the table examples, which are a bit
hairy and written by John Gill, so I put it on the back burner.  For
barh, however, I did make the y axis mean the center, since we are all
in agreement that this is the more intuitive behavior.


JDH

Thanks for the reply. Surprises me that people would really argue about this 
issue, but well. :-)

Attached is a slighly changed patch: effectively, it does not change the 
layering any more, but it changes the internals of errorbar() in a way, that 
allows to swap the layering by simply swapping two blocks of code. (Which was 
nontrivial before, since the color from the plot line can only be read after 
it has been drawn.)

Whether this swapping is then done by hand, by an option or an entry in the rc 
file can be left to the future.

Maybe, submitting this patch will save someone else the time to figure it out 
himself.

Ciao,
Nobbi


On Friday 22 October 2004 12:27, gary ruben wrote:
> Hi Norbert,
> I think this is the best place to post patches like this.
> There was some discussion a while back about the layer ordering of
> errorbars and other objects in the plot window. It was decided to leave it
> unchanged for the moment since about half of users want it one way and half
> the other. Some ideas were raised like being able to specify layers as
> command parameters or change the default in the resource file. I think that
> until this is pursued, it is unlikely that anything will be done.
> Gary
>
> *********** REPLY SEPARATOR  ***********
>
> On 22/10/2004 at 11:29 Norbert Nemec wrote:
> > Hi there,
> >
> > here is a small patch layering errorbars behind the line of the plot
> > itself.
> >
> > Rationale:
> > If you use ecolor and have a plot with many dense points, the errorbars
> > may be
> > covering the line itself completely. The relayering makes the line fully
> > visible and puts the errorbars as "additional information" in the
> > background.
> >
> > Greetings,
> > Norbert
> >
> > PS: If sending small patches to this mailing list is a suboptimal way of
> > submission, please tell me how to do it better.
> >
> > --
> > _________________________________________Norbert Nemec
> >          Bernhardstr. 2 ... D-93053 Regensburg
> >      Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
> >            eMail: <No...@Ne...>
> >
> >
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> > Use IT products in your business? Tell us what you think of them. Give us
> > Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out
> > more
> > http://productguide.itmanagersjournal.com/guidepromo.tmpl
> > _______________________________________________
> > Matplotlib-devel mailing list
> > Mat...@li...
> > https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
>
> ------------------------------------
> Gary Ruben     gr...@bi...
> <http://users.bigpond.net.au/gazzar>
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> Use IT products in your business? Tell us what you think of them. Give us
> Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
> http://productguide.itmanagersjournal.com/guidepromo.tmpl
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel

-- 
_________________________________________Norbert Nemec
         Bernhardstr. 2 ... D-93053 Regensburg
     Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
           eMail: <No...@Ne...>

Hi Norbert,
I think this is the best place to post patches like this.
There was some discussion a while back about the layer ordering of
errorbars and other objects in the plot window. It was decided to leave it
unchanged for the moment since about half of users want it one way and half
the other. Some ideas were raised like being able to specify layers as
command parameters or change the default in the resource file. I think that
until this is pursued, it is unlikely that anything will be done.
Gary

*********** REPLY SEPARATOR  ***********

On 22/10/2004 at 11:29 Norbert Nemec wrote:

> Hi there,
> 
> here is a small patch layering errorbars behind the line of the plot
> itself. 
> 
> Rationale:
> If you use ecolor and have a plot with many dense points, the errorbars
> may be 
> covering the line itself completely. The relayering makes the line fully 
> visible and puts the errorbars as "additional information" in the
> background.
> 
> Greetings,
> Norbert
> 
> PS: If sending small patches to this mailing list is a suboptimal way of 
> submission, please tell me how to do it better.
> 
> -- 
> _________________________________________Norbert Nemec
>          Bernhardstr. 2 ... D-93053 Regensburg
>      Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
>            eMail: <No...@Ne...>
> 
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> Use IT products in your business? Tell us what you think of them. Give us
> Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out
> more
> http://productguide.itmanagersjournal.com/guidepromo.tmpl
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel


------------------------------------
Gary Ruben     gr...@bi...
<http://users.bigpond.net.au/gazzar>

Hi there,

on one of my machines, matplotlib.matlab segfaults on import. On another 
machine, it works fine. I already traced the segfault to the library 
_na_transforms.so, but now I don't know how to continue from here. Can anyone 
give me a hint how to debug a python module written in C/C++?

I have plenty of experience in C/C++, but I'm just starting on python.

Thanks for help!

Ciao,
Norbert

-- 
_________________________________________Norbert Nemec
         Bernhardstr. 2 ... D-93053 Regensburg
     Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
           eMail: <No...@Ne...>

Hi there,

here is a small patch layering errorbars behind the line of the plot itself. 

Rationale:
If you use ecolor and have a plot with many dense points, the errorbars may be 
covering the line itself completely. The relayering makes the line fully 
visible and puts the errorbars as "additional information" in the background.

Greetings,
Norbert

PS: If sending small patches to this mailing list is a suboptimal way of 
submission, please tell me how to do it better.

-- 
_________________________________________Norbert Nemec
         Bernhardstr. 2 ... D-93053 Regensburg
     Tel: 0941 - 2009638 ... Mobil: 0179 - 7475199
           eMail: <No...@Ne...>