Just a couple of quick notes:

1. Is it worth adding this to the regular tkinter module? I suspect it would be useful for applications other than IDLE (I would personally make great use of such a method). We would have to ensure it is properly documented so people know when / not to use it.

2. I suspect we would need to add a flag to ensure the user doesn't / cannot try to create a new Tk instance after calling the Tcl_Finalize wrapper (i.e. an exception would be raised by _tkinter.create)

I'm suggesting this as an immediate fix for the reported issue as it relates to IDLE.

Making this more generally available would require figuring out how it best fits into tkinter's set of abstractions, documenting it, testing it thoroughly, etc. I'm all for it, but IMO we should open a separate issue for that and not have it further delay this fix for IDLE.

I'm looking into the test failure. Surprisingly, it is consistent and I can reproduce it locally on Windows 10.

When you're done making the requested changes, leave the comment: I have made the requested changes; please review again.

I'm looking into the test failure. Surprisingly, it is consistent and I can reproduce it locally on Windows 10.

It appears that merging in the main branch fixed the failing tests. 🤷

I have made the requested changes; please review again.

Thanks for making the requested changes!

@terryjreedy: please review the changes made to this pull request.

I rebuilt branch and help(_tkinter._finalize_tcl) properly displays the new docstring. Once the blurb is fixed, I'm okay with you merging when you are.

Note: This does not work on more up-to-date Linux (e.g. Arch, Debian testing). It's a Tk issue (idk why).

Note: This does not work on more up-to-date Linux (e.g. Arch, Debian testing). It's a Tk issue (idk why).

Could you elaborate please? What exactly doesn't work, and in what way?

On the two distros named (I haven't tested on e.g. Fedora), Tk clipboard content is cleared on exit regardless of the new call. I proved it was a Tk issue because the following also results in an empty clipboard:

>>> wish
% clipboard clear
% clipboard append {testing 123}
% exit

On the two distros named (I haven't tested on e.g. Fedora), Tk clipboard content is cleared on exit regardless of the new call. I proved it was a Tk issue because the following also results in an empty clipboard:

>>> wish
% clipboard clear
% clipboard append {testing 123}
% exit

Oh, then that's different from what we've been trying to fix up until now, since on all platforms tested so far the above worked as expected; it only didn't work via tkinter.

That should be reported as a Tcl/Tk bug, and shouldn't affect this issue and PR.

@serhiy-storchaka, with the latest change based on your review comment, do you think this is good to go?

This PR is stale because it has been open for 30 days with no activity.

@serhiy-storchaka are we good to merge this?

I just had Python segfault while running IDLE with this change. Let's hold off on merging for now.

This missed the boat for inclusion in Python 3.9 which accepts security fixes only as of today.

Note: This does not work on more up-to-date Linux (e.g. Arch, Debian testing). It's a Tk issue (idk why).

The way the clipboard works on X11, is that it only stores a reference to the data when copying, and only retrieves the actual data when a paste is requested. Thus, the source application must provide the data at paste time, otherwise the clipboard will be empty.

The clipboard content will only persist on X11 if there's a clipboard manager running, that constantly "harvests" clipboard selections.

I just had Python segfault while running IDLE with this change. Let's hold off on merging for now.

I smoke-tested (python -m idlelib-select-Ctrl+C-close) IDLE from the head of this PR and couldn't reproduce the crash. @taleinat could you recheck if the underlying reason was fixed during last two years, please?

All commit authors signed the Contributor License Agreement.