MNT: Unpin pyparsing, xfail error message tests for pyparsing 3.1.0 #26431
Conversation
|
I did run locally with pyparsing 3.1.0 to ensure that the tests xfailed as desired. CI does not have anything that would pin that specific version, so not tested automatically. I initially ran only the mathtext subset, found that there are two more in |
May require backporting a portion of matplotlib#26198 to the 3.7 branch if we wish to have a 3.7.3. (specifically the part where token, placeable, and auto_delim are excluded from setName)
|
Labeling as 3.7.3. While the backport shouldn't fail, we probably do want to include the lines from #26198 in there. |
| @@ -19,7 +19,7 @@ dependencies: | |||
| - pillow>=6.2 | |||
| - pybind11>=2.6.0 | |||
| - pygobject | |||
| - pyparsing!=3.1.0 | |||
| - pyparsing>=2.3.1 | |||
There was a problem hiding this comment.
Do you not need to exclude the version that doesn't work?
|
Given that we have the xfails on the tests and functionally the issue is that error messages on badly formatted strings (not failing for correctly formatted strings) I think it is OK to not put in any exclusionary pins. |
…ge tests for pyparsing 3.1.0
…431-on-v3.7.x Backport PR #26431 on branch v3.7.x (MNT: Unpin pyparsing, xfail error message tests for pyparsing 3.1.0)
PR summary
Closes #26152
May require backporting a portion of #26198 to the 3.7 branch if we wish to have a 3.7.3.
(specifically the part where token, placeable, and auto_delim are excluded from
setName)PR checklist