Re: bug#32029: [PATCH] xref-find-definitions-at-mouse

[Dmitry Gutov <dgutov@yandex.ru>]

On 7/4/18 4:02 PM, Tobias Gerdin wrote:

> Well, if we want the pushed mark to correspond to only the location of 
> the actual identifier it's easy, just call `mouse-set-point` and then 
> xref-find-definitions interactively and you're good. But then we throw 
> away the users actual location before the click which I don't think is 
> nice, but just my opinion.

I'm not sure we want that. What's more, capturing the previous state 
will become more difficult, as it will include window configuration (a 
different window, or frame, may have been selected before the mouse click).

>>
>> Regarding the patch, I'm fine with it, but I'd like someone else to 
>> confirm that the name of the new function makes sense.
> The name was chosen to match `ffap-at-mouse`. Would be good to have a 
> second opinion on that.
> 
> Would you like me to draft a NEWS entry? In particular I think it should 
> draw attention to the need for the user to bind a mouse event to the 
> command in order to work.

Yes, please. Also, please keep the mailing list in Cc.

> Personally I would like to have a default 
> binding for this. For comparison `ffap-bindings` binds `ffap-at-mouse` 
> to S-mouse-3. I could initiate a discussion on emacs-devel about it. In 
> the best of worlds we would have a function bound to some key which 
> unifies ffap and xref (invoking the former if the thing at point looks 
> like a file or URL) which would expose both functionalities by default.

If you feel the need, please go ahead and start the discussion.

> Btw, in previous Emacsen one used to be able to hit M-v in order to 
> focus the completions buffer window to select completion (if there is 
> more than one matching completion). This no longer seems to work for me 
> in recent Emacs, is that by design? (I suspect this was the 
> `switch-to-completions` command.)

It might be a regression. Try reporting it as a bug.