[Matplotlib-devel] Phone call notes 2017-08-07

Thomas Caswell tcaswell at gmail.com
Mon Aug 14 17:02:41 EDT 2017


Folks,

Sorry these are so late getting out.  Myself, Eric Firing, and Ryan May
where on the call.


 - talked about image interpolation PR, I have a todo list in a comment on
that one still
 - talked about unbreaking %matplotlib notebook
   - source of breaking is that it got upgraded to use ipywidegts, but v4
   - there have been many breaking changes between than and now
   - plan is to revert that PR and go back to injecting our own JS
     (because it works across notebook versions)
 - talked about long term fate of 'classic' mode
   - when we have to re-generate images for other reasons, switch those
tests to using the new default
 - discussed (and merged) #8949
 - discussed #8944
 - discussed #9002 and hi-dpi
   - it is a mess!
   - Eric's summary of what is going on
     - at rendering time, there is a 'true dpi' depending on renderer
       and what it is rendering too (what screen)
     - best case user only knows about 'physical' units
     - our renderers need to do best to go out pixelated output
       - in vector backends we are doing an intermediate rendering to their
represenation
     - in Agg we are doing both steps ourselves
     - maybe we should think of this as two stage process to get everything
to physical units and
       the second step
     - maybe we should add anothre layer to the transform stack

Tom
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/matplotlib-devel/attachments/20170814/9c9e2f81/attachment.html>


More information about the Matplotlib-devel mailing list