Repository navigation
Backport eps work - #17257
Backport eps work#17257QuLogic merged 4 commits into
Conversation
Sorry, something went wrong.
Define \mathdefault as a noop in the usetex preamble.
Conflicts:
lib/matplotlib/ticker.py
change of a near-by conditional from np.abs -> abs caused
conflict
Fix use of psfrags in ps backend + usetex.
Conflicts:
lib/matplotlib/backends/backend_ps.py
- near by change of rcParams -> mpl.rcParams
lib/matplotlib/texmanager.py
- do not implicitly backport rcParam deprecations
Fix baseline alignment when using usetex.
Conflicts:
lib/matplotlib/tests/test_usetex.py
- import np and formatting issues on decorator
FIX: eps + usetex combo.
anntzer
left a comment
There was a problem hiding this comment.
postci.
Sorry, something went wrong.
|
The reason I didn't write a test is that ghostscript actually doesn't error when rendering the "bad" ps files, just gives a blank page, and as usual I didn't want to add a baseline image. I guess one could use the usual trick of rendering the thing, "manually" (i.e. by calling ghostscript in the test) converting to png, and checking that the page isn't all white... |
Sorry, something went wrong.
|
I went through a very similar thought process. Would it be reliable to render the example @QuLogic came up with verify that the mediabox is correct? |
Sorry, something went wrong.
|
I guess "yes", but it's really a super indirect test, it wouldn't be obvious at all why these values are the correct ones. |
Sorry, something went wrong.
QuLogic
left a comment
There was a problem hiding this comment.
Seems reasonable.
Sorry, something went wrong.
PR Summary
This is the stack of PRs that I had to backport to make them all apply relatively cleanly. All are arguably bug-fixes. The 2nd (#16913) and 4th (#17255 ) are the likely the critical ones.
I have verified that the example from #17253 works. Probably should add that as a test...
(re) closes #17253.