advise read-key as well as read-key-sequence

[willghatch]

As discussed in #857 and #858, the use of read-key in the evil-surround package broke evil-repeat, or .. This is because read-key changes the result of this-command-keys just like read-key-sequence. read-key-sequence is advised to fix this issue. This patch merely gives the same advice to read-key, fixing the issues that have been caused by its use.

[ninrod]

hum. 1 test case failed. can't find it.

[wasamasa]

In graphical Emacs, this fails two repeat-related tests, evil-test-repeat-digraph and evil-test-repeat-kill-buffer.

[willghatch]

Ah, when I ran the tests I think I didn't see the failure because it was picking up .elc files. I've now fixed the tests and I get no errors in terminal or graphical mode. But having fixed the test, I'm now not sure that the advice for read-key-sequence (and with this patch also read-key) is actually correct. As is, it calls any function that is the result of (evil-repeat-type this-command t), and when that is evil-repeat-keys that makes a lot of sense. But when it is evil-repeat-changes (which is what evil-insert-digraph uses), it makes a lot less sense to me. I think maybe the advice should only call it when it is evil-repeat-keys.

[willghatch]

Huh, I get no errors on my machine, but travis is getting one and I can't tell what it is that failed. If I understand the travis page right, it's passing for emacs 24.3 but failing for all the others it's trying. I'm on emacs 26 built from emacs git master of a couple months ago... Maybe tomorrow I'll get a version of emacs it's failing on and test it there.

[wasamasa]

This is unfortunate indeed. IIRC @TheBB is working on a version of the tests where as much as possible runs in batch mode.

[TheBB]

Yeah I am, they're so damn fickle though.

[TheBB]

If anyone is still interested in this, I rebased it on master (with a more sensible travis setup) and there's only one failure for 24.5.

[willghatch]

On Fri, Nov 22, 2019 at 06:28:02AM -0800, Eivind Fonn wrote: If anyone is still interested in this, I rebased it on master (with a more sensible travis setup) and there's only one failure for 24.5.

Cool. I still care (to get evil-surround to work properly with multi-byte utf-8 characters, which I frequently want), but unfortunately emacs improvements are pretty low on my priority list right now. So I don't really have time to work on this in the near future.

[wbolster]

also still interested since evil-swap-keys.el and evil's repeat facility seem to interfere and it seems like this is related