FIX: don't close figures if switch_backend is a no-op#14471
FIX: don't close figures if switch_backend is a no-op#14471anntzer merged 3 commits intomatplotlib:mainfrom
Conversation
anntzer
left a comment
There was a problem hiding this comment.
modulo #14426 (comment) but heh
9bb75fe to
02d3c1b
Compare
b4f9ebb to
3eeb476
Compare
dopplershift
left a comment
There was a problem hiding this comment.
LGTM assuming CI passes.
30ebdf9 to
78ca4cb
Compare
|
rebased, but there is something funny going on here, pushed to 3.3. |
78ca4cb to
5ee99ae
Compare
5ee99ae to
a7e7cf1
Compare
a7e7cf1 to
5ebb5f9
Compare
|
Can we have some documentation on the behavior; maybe in the |
|
This took us just 6 releases. 😮 |
|
@anntzer can you weigh in on this again so Tom doesn't have to rebase any more? |
|
I'm still not really convinced it's the right approach (per my comment above), also I'd say that if we really want to go that route we should skip closing whenever the event loops are compatible (e.g. going from qtagg to qtcairo), not just when the backends are exactly equal. OTOH we can always decide on that later and I don't feel overly strongly about this. |
I think this should go in for 3.1.1 as it helps to reduce the
action-at-a-distance issues around the auto-backend / IPython
configuration.