e.g.plot([1, 2], ls=(None, (4, 4))) which was previously a synonym
forplot([1, 2], ls=(0, (4, 4))) which means "dash-pattern of
4pt-long dashes separated by 4pt spaces, with an offset of 0pt at
start".

Passing None instead of 0 was already not (likely, never) supported for
pdf, ps, or svg (an exception is raised at savefig() time for ps/svg, an
invalid pdf is generated). There isn't much of a point in supporting
None (this also makes e.g. mplcairo more complex because of static
typing in C++ extensions), so just deprecate it.

Goes on top of#15827.
Edit:closes#14996, which is basically the issue described here.

PR Checklist

Has Pytest style unit tests
Code isFlake 8 compliant
New features are documented, with examples if plot related
Documentation is sphinx and numpydoc compliant
Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way

anntzer added the status: waiting for other PR label

Dec 4, 2019

anntzer force-pushed thedashoffset-none branch 2 times, most recently from87e9acf to9300485Compare

December 5, 2019 01:02

tacaswell added this to thev3.3.0 milestone

Dec 5, 2019

Copy link

Member

tacaswell commentedDec 5, 2019

Could we normalizeNone as input to 0 instead?

The failures look real, but did not dig into them....

anntzer added the status: work in progress label

Dec 5, 2019

Copy link

ContributorAuthor

anntzer commentedDec 5, 2019

The failures are real, need to look into it.
I think the (small) cleanup to Line2D from not normalizing (allowing to remove a few checks for None) is also nice to have. Again note that this has (likely) never worked for pdf/ps/svg so I doubt there are many people who rely on it...

anntzer force-pushed thedashoffset-none branch from9300485 to5345d49Compare

December 5, 2019 14:10

anntzer force-pushed thedashoffset-none branch from5345d49 to5b1be65Compare

January 21, 2020 22:29

anntzer mentioned this pull request

Jan 21, 2020

format error saving eps figure using custom linestyle#14996

Closed

anntzer removed the status: waiting for other PR label

Jan 24, 2020

anntzer force-pushed thedashoffset-none branch 2 times, most recently from838c6fc to93f43a3Compare

January 26, 2020 17:15

anntzer removed the status: work in progress label

Jan 26, 2020

Copy link

ContributorAuthor

anntzer commentedJan 26, 2020

figured out the failure, now fixed, should be good to go.

timhoffm approved these changes

Jan 26, 2020

View reviewed changes

QuLogic added the status: needs rebase label

Mar 10, 2020

QuLogic approved these changes

Mar 10, 2020

View reviewed changes

Copy link

Member

QuLogic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

Rebase and maybe small tweak to a sentence.

doc/api/next_api_changes/deprecations.rst OutdatedShow resolvedHide resolved

Deprecate support for dash-offset = None.

eb76378

e.g. `plot([1, 2], ls=(None, (4, 4)))` which was previously a synonymfor `plot([1, 2], ls=(0, (4, 4)))` which means "dash-pattern of4pt-long dashes separated by 4pt spaces, with an offset of 0pt atstart".Passing None instead of 0 was already not (likely, never) supported forpdf, ps, or svg (an exception is raised at savefig() time for ps/svg, aninvalid pdf is generated).  There isn't much of a point in supportingNone (this also makes e.g. mplcairo more complex because of statictyping in C++ extensions), so just deprecate it.

anntzer force-pushed thedashoffset-none branch from93f43a3 toeb76378Compare

March 10, 2020 03:52

Copy link

ContributorAuthor