[3.13] gh-124111: Update tkinter for compatibility with Tcl/Tk 9.0.0 (GH-124156) #127364
Conversation
…ythonGH-124156) (cherry picked from commit 47cbf03)
|
The biggest part of this change are tests. |
| if isinstance(val, tuple) and len(val) == 0: | ||
| return '' |
There was a problem hiding this comment.
I'm not sure if this can affect existing code.
There was a problem hiding this comment.
This change is a backport from main. If the code doesn't make sense, it should be changed in main first.
There was a problem hiding this comment.
Maybe a review from someone who's not a tk expert at all would still be helpful?
Most of the PR is changes to tests, which look reasonable, and should still do their job after this reorganization.
I added comments for the two behaviour changes.
I can't comment on Windows packaging. Unfortunately, @zooba's review on the original PR looks unfinished.
| @@ -637,6 +638,7 @@ Tkapp_New(const char *screenName, const char *className, | |||
| v->ListType = Tcl_GetObjType("list"); | |||
| v->StringType = Tcl_GetObjType("string"); | |||
| v->UTF32StringType = Tcl_GetObjType("utf32string"); | |||
| v->PixelType = Tcl_GetObjType("pixel"); | |||
There was a problem hiding this comment.
This is NULL if "pixel" is not registered. I assume that a NULL typePtr isn't valid, so the later use is OK.
There was a problem hiding this comment.
This change is a backport from main. I suggest to enhance the code in the main branch first.
|
The changes that are there shouldn't impact the Windows builds with Tcl/Tk 8, but without installer changes we aren't able to release with Tcl/Tk 9. IIRC, that wasn't the intent of the original PR (someone wanted to be able to test their own builds), and we can't update 3.13 to Tcl/Tk 9 anyway, so it doesn't particularly matter. |
|
Downstream distributors and users will want to move Tk 9 if for no other reason than it is new. This is already happening, i.e. with Homebrew for macOS. So it makes sense that we should support Tk 9 in 3.13.x. That said, Tk 9.0.0 was just released 2 months ago and the Tcl/Tk project is working on a 9.0.1 release for the near future so I don't think we need to be in a big hurry to move everything to Tk 9. (I have not yet done any building or testing of tkinter with Tk 9 on macOS.) As far as additional review of the tkinter changes, @serhiy-storchaka wrote most of the existing tkinter tests and made most of the changes to tkinter in recent years so would be my first recommendation for additional review. I don't know that we have had any substantial feedback from 3.14 alpha users who might be using Tk 9 in their own builds. FWIW, the original 3.14 PR was submitted by the Tk core developer who primarily supports Tk on macOS. |
|
Just to be clear, I do not think we should consider making the 3.13 installers use Tk 9. The question is really whether the non-test changes necessary to make it possible to use Tk 9 are simple and non-invasive enough to backport it to 3.13. |
|
I have doubts about handling the pixel type. This is a new feature, and there may be better representation of it than str in Python. You can merge backports now if this is needed to fix test failures, but I am planning to revert some changes and replace them with less invasive test-only changes. |
|
I'm sorry but I don't understand well the feedback on this change. Do you want to change to land into 3.13? If not, why not? Should I remove the PCbuild/ changes? |
I'm not worried about them. As I think I said elsewhere, it's enough to build and test, but not enough to generate our installers with the updated files (and I'd backport those changes if/when we do them if someone wants them). |
|
Since nobody seems to be eager to review/approve this PR, I plan to abandon it. |
|
I abandon my PR because of the lack of interest to get it merged. |
Simple backport from main to 3.13 branch with no merge conflict.
(cherry picked from commit 47cbf03)