Improving axis spine joins and final tick placement

Hi all,

I have a question/request about improving a minor aspect of plot
cosmetics, namely that axis spines seem to be rendered as 4 individual
lines, meaning that the corners of the axes boundary do not match up
with a neat corner but have a "stair" appearance. I've attached a
screenshot from a zoomed PDF showing this effect, made with MPL 1.3.1
from Ubuntu 14.04... so apologies if it's already fixed in the latest
1.4.x versions. This effect is also visible in PNG output, as just a
couple of pixels in the corners that look "ragged". Would it be possible
to tidy this up... or point me to where in the code this rendering is
done, if it's something easy that I could maybe help with?

At the same time, I note in this zoom that the axis is showing tick
marks at the very end(s) of the axis, where they overlap with the other
axis/plot boundary line: is there an automatic way to elide tick marks
in that redundant position?

Apologies if this isn't a good place for these queries/requests -- I had
a look at the matplotlib-user and -announce list archives linked from
the web page and they seem to have gone defunct in 2012, hence coming
here. I am happy to do a bit of development to address little cosmetic
tweaks like this, but am not yet familiar with MPL internals.

Thanks!
Andy

PS. I also have a question about how to enable old-style figures in a
font when using the TeX/PGF rendering backed, cf.
\usepackage[osf]{mathpazo}, but I'll wait to see if this is an
appropriate place for such questions before troubling you with that!

mpl-spinejoins-and-finalticks.png

···

--
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN

Andy,

The issue of the axes frame corners should be fixed in 1.4

The user mailing list is still a going concern, the SF archives are just broken. They know (and the archive pages are less broken than they used to be), but have not fixed it yet.

Tom

···

On Sun, Oct 26, 2014 at 11:50 AM, Andy Buckley <andy@…55…1224…> wrote:

Hi all,

I have a question/request about improving a minor aspect of plot

cosmetics, namely that axis spines seem to be rendered as 4 individual

lines, meaning that the corners of the axes boundary do not match up

with a neat corner but have a “stair” appearance. I’ve attached a

screenshot from a zoomed PDF showing this effect, made with MPL 1.3.1

from Ubuntu 14.04… so apologies if it’s already fixed in the latest

1.4.x versions. This effect is also visible in PNG output, as just a

couple of pixels in the corners that look “ragged”. Would it be possible

to tidy this up… or point me to where in the code this rendering is

done, if it’s something easy that I could maybe help with?

At the same time, I note in this zoom that the axis is showing tick

marks at the very end(s) of the axis, where they overlap with the other

axis/plot boundary line: is there an automatic way to elide tick marks

in that redundant position?

Apologies if this isn’t a good place for these queries/requests – I had

a look at the matplotlib-user and -announce list archives linked from

the web page and they seem to have gone defunct in 2012, hence coming

here. I am happy to do a bit of development to address little cosmetic

tweaks like this, but am not yet familiar with MPL internals.

Thanks!

Andy

PS. I also have a question about how to enable old-style figures in a

font when using the TeX/PGF rendering backed, cf.

\usepackage[osf]{mathpazo}, but I’ll wait to see if this is an

appropriate place for such questions before troubling you with that!

Dr Andy Buckley, Royal Society University Research Fellow

Particle Physics Expt Group, University of Glasgow / PH Dept, CERN



Matplotlib-devel mailing list

Matplotlib-devel@lists.sourceforge.net

https://lists.sourceforge.net/lists/listinfo/matplotlib-devel


Thomas Caswell
tcaswell@…149…

Thanks Thomas, that's great to know. I'll try using a newer version from
Github, and nice that the correction will gradually propagate to our users.

I'll ask my other questions on the user list; thanks for explaining!

Andy

···

On 26/10/14 16:18, Thomas Caswell wrote:

Andy,

The issue of the axes frame corners _should_ be fixed in 1.4

The user mailing list is still a going concern, the SF archives are just
broken. They know (and the archive pages are _less_ broken than they
used to be), but have not fixed it yet.

Tom

On Sun, Oct 26, 2014 at 11:50 AM, Andy Buckley <andy@…1224… > <mailto:andy@…1224…>> wrote:

    Hi all,

    I have a question/request about improving a minor aspect of plot
    cosmetics, namely that axis spines seem to be rendered as 4 individual
    lines, meaning that the corners of the axes boundary do not match up
    with a neat corner but have a "stair" appearance. I've attached a
    screenshot from a zoomed PDF showing this effect, made with MPL 1.3.1
    from Ubuntu 14.04... so apologies if it's already fixed in the latest
    1.4.x versions. This effect is also visible in PNG output, as just a
    couple of pixels in the corners that look "ragged". Would it be possible
    to tidy this up... or point me to where in the code this rendering is
    done, if it's something easy that I could maybe help with?

    At the same time, I note in this zoom that the axis is showing tick
    marks at the very end(s) of the axis, where they overlap with the other
    axis/plot boundary line: is there an automatic way to elide tick marks
    in that redundant position?

    Apologies if this isn't a good place for these queries/requests -- I had
    a look at the matplotlib-user and -announce list archives linked from
    the web page and they seem to have gone defunct in 2012, hence coming
    here. I am happy to do a bit of development to address little cosmetic
    tweaks like this, but am not yet familiar with MPL internals.

    Thanks!
    Andy

    PS. I also have a question about how to enable old-style figures in a
    font when using the TeX/PGF rendering backed, cf.
    \usepackage[osf]{mathpazo}, but I'll wait to see if this is an
    appropriate place for such questions before troubling you with that!

    --
    Dr Andy Buckley, Royal Society University Research Fellow
    Particle Physics Expt Group, University of Glasgow / PH Dept, CERN

    ------------------------------------------------------------------------------

    _______________________________________________
    Matplotlib-devel mailing list
    Matplotlib-devel@lists.sourceforge.net
    <mailto:Matplotlib-devel@lists.sourceforge.net>
    https://lists.sourceforge.net/lists/listinfo/matplotlib-devel


Thomas Caswell
tcaswell@…149… <mailto:tcaswell@…149…>

--
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN