[Matplotlib-devel] Help with constrained_layout

[Jody Klymak]

Hi all,

I know a lot of bandwidth has been taken up w/ the 2.1 release, but that 
seems to be settling down now. I’ve been working hard on the 
constraint-driven layout.  This is an alternate to `tight_layout`, and 
based on [kiwi-solver](https://github.com/nucleic/kiwi/tree/master/py)  
I think its mostly feature-complete at this point, and handles numerous 
situations not handled by `tight_layout` like `suptitle`, over-spilling 
legends, colorbars that belong to a group of axes (as opposed to just 
one axis), nested gridspecs, etc.  (Conversely, it doesn’t work at all 
w/ `plt.subplot` or `plt.subplot2grid` because each call to these 
functions makes their own gridspec, and we do the layout within 
individual gridspecs).

I’m writing to appeal for some help testing this, or coming up with 
situations that I didn’t think about.  There is one request for 
AnchoredOffsetBoxes, which I haven’t done yet, but I think that is a 
0.2 feature once folks think the basic idea is OK. I’m also more than 
open to addressing any implementation concerns/suggestions.

Right now, automatic installation only works with python 3.6 because we 
require kiwi-solver gitub master version for python 2.7.  But you should 
be able to install with `pip install .` into a virtualenv.

https://github.com/matplotlib/matplotlib/pull/9082

There is also a guide in 
`tutorials/intermediate/constrainedlayout_guide.py`.  It probably still 
has some rough edges, but suggestions welcome.

Thanks for any help.

Cheers,   Jody
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-devel/attachments/20170929/c5804853/attachment-0001.html>