On 16 Aug 2019, Juri Linkov wrote:
>>> Should your new command catch this error?
>>
>> It could.  I'm not sure exacly what to do with the error, though.
>> The new version below displays a brief message and then returns to the
>> isearch prompt:
>>
>> This feels klugey, but I'm not sure what a better way is.  Do you have
>> any ideas?
>
>Isearch used to catch errors and display the error message
>at the end of the isearch message by setting `isearch-error'.
>Why it doesn't do this now, I don't know, this needs more investigation.
>Maybe just setting `isearch-error' to the error message would help.

Well, I tried setting `isearch-error' in the error case -- instead of using `message' as the current patch does -- but it didn't seem to have any useful effect.  I just got the "I-search:" prompt back immediately, with no indication that the character I had typed was not found.  Then I tried keeping the current `message' code and *also* setting `isearch-error', but that seemed to have no effect either.  The result was exactly the same as the current message behavior.

So I've left the code as-is; the current patch is attached here.

Do we have consensus to install it?

I don't have a good sense of whether there's consensus yet.  We had some discussion about what keybinding to use within isearch, but (I think) the outcome was that M-C-c was an okay choice.  We also seemed to agree that the current way of handling the "char not found" error is a bit klugey, but we don't know a better way to do it (or at least I don't know a better way).

(Through the discussion, thanks to you Juri, I also learned about the more general `isearch-yank-on-move'.  For those who deliberately choose that behavior and won't be confused by it, it's pretty neat!  But I think it's orthogonal to this patch, since the point of this patch is to add a transparent new behavior within traditional isearch -- that is, a behavior that won't get in the way or cause any surprises for those who don't learn about it.)

Best regards,
-Karl