@terryjreedy, please take a look. I tested this change and fixed it. It LGTM now.

1. loadfile twice calls stat, saving to temp local and thence to self.xxx. Is there an advantage to doing the stat call while the file is open? (Faster?)

2. One of four stat calls is wrapped in try-except. Should all 4 be so wrapped? Or at least the other save call?

3. In save, the comparison != might just as well be <. See 4.

4. The default return for SaveAs and SaveCopyAs is the current file. So I think the stat call and check should either be moved to writefile or put in a new method or function. Not sure of the best option yet.

1. I I initially was going to use os.fstat() (it is safer), but it cannot be used in other two places, so I leave it for now. Only one of them is called, the second way is executed if the first one fails. It is also more reliable to call here, in try/except blocks.
2. This one is only can easily fail. All three others are very unlikely to fail, because they are called immediately after reading or writing a file. The two in loadfile are also in try/except blocks.
3. I also thought about this, but it is better to keep !=. Imagine that you restored the earlier version of the file from archive or backup.
4. SaveAs updates the current file name and the timestamp. SaveCopyAs keeps the current file name and the timestamp. Both call writefile. Initially updating the timestamp was in set_filename, but I separated them, because it failed if set_filename was called with the name of non-existing file (see gh-50948: IDLE: Warn if saving a file will overwrite a newer version #17578 (review)).

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

Thanks @ZackerySpytz for the PR, and @serhiy-storchaka for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13, 3.14, 3.15.
🐍🍒⛏🤖

GH-151026 is a backport of this pull request to the 3.15 branch.

GH-151027 is a backport of this pull request to the 3.14 branch.

GH-151028 is a backport of this pull request to the 3.13 branch.