Re: [Emacs-diffs] master 618310c: Work on `tramp-completion-mode-p'

Re: [Emacs-diffs] master 618310c: Work on `tramp-completion-mode-p'

[Dmitry Gutov]

On 26.02.2017 13:45, Michael Albinus wrote:

>   (progn (defun tramp-completion-mode-p ()
>     "Check, whether method / user name / host name completion is active."
>     (or
>      ;; Signal from outside.  `non-essential' has been introduced in Emacs 24.
>      (and (boundp 'non-essential) (symbol-value 'non-essential))
> -   tramp-completion-mode)))
> +   ;; This variable has been obsoleted in Emacs 26.
> +   tramp-completion-mode
> +   ;; Fallback.  Some completion packages still don't support
> +   ;; `non-essential' sufficiently.

Any packages in particular? Should we try working with their authors?

> +   (equal last-input-event 'tab))))

Not sure I understand what's going on here. It seems you are making 
`non-essential' being non-nil and `tab' being the last input event have 
the same impact.

Whereas, IME, the user pressing TAB usually means that the completion 
operation *is essential*, and they would probably be willing to input 
their connection credentials if necessary.

Like the discussion in https://debbugs.gnu.org/25607 has touched on.

Re: [Emacs-diffs] master 618310c: Work on `tramp-completion-mode-p'

[Michael Albinus]

Dmitry Gutov <dgutov@yandex.ru> writes:

Hi Dmitry,

>> +   ;; Fallback.  Some completion packages still don't support
>> +   ;; `non-essential' sufficiently.
>
> Any packages in particular? Should we try working with their authors?

Yes, I'll intend to compile a list of problems, and go through then.

>> +   (equal last-input-event 'tab))))
>
> Not sure I understand what's going on here.

I've reintroduced a very old workaround in Tramp, which is there since
2002 or so. Recently, I've removed it (commit
611088642fa60f4e29ccaac0236d2bac3c10e98f from Feb 15), because I thought
Tramp doesn't need it any longer. But that was too early; there are
still problems then, and I don't want to disappoint users of Emacs
master. So I've taken this back.

The plan is still to get rid of it for Emacs 26.

Best regards, Michael.

Re: [Emacs-diffs] master 618310c: Work on `tramp-completion-mode-p'

[Dmitry Gutov]

On 28.02.2017 11:05, Michael Albinus wrote:

> I've reintroduced a very old workaround in Tramp, which is there since
> 2002 or so. Recently, I've removed it (commit
> 611088642fa60f4e29ccaac0236d2bac3c10e98f from Feb 15),

I think you mean adfb6f1dc26a927cf4bab24bdfae54b51e3ae0ec.

> because I thought
> Tramp doesn't need it any longer. But that was too early; there are
> still problems then, and I don't want to disappoint users of Emacs
> master. So I've taken this back.
> 
> The plan is still to get rid of it for Emacs 26.

I was commenting on the details of that workaround. But if you prefer to 
consider all that later, that's fine.

Re: [Emacs-diffs] master 618310c: Work on `tramp-completion-mode-p'

[Michael Albinus]

Dmitry Gutov <dgutov@yandex.ru> writes:

> On 28.02.2017 11:05, Michael Albinus wrote:
>
>> I've reintroduced a very old workaround in Tramp, which is there since
>> 2002 or so. Recently, I've removed it (commit
>> 611088642fa60f4e29ccaac0236d2bac3c10e98f from Feb 15),
>
> I think you mean adfb6f1dc26a927cf4bab24bdfae54b51e3ae0ec.

Indeed, sorry. I've mentioned the related commit in the Tramp repo.

> I was commenting on the details of that workaround. But if you prefer
> to consider all that later, that's fine.

I want to get rid of this workaround. That's why it's wasted time to
discuss its pros and cons, I believe.

But thanks that you have taken the time to chime in! I'm pretty sure
there will be lot of opportunities next time to speak about Tramp file
name completion ...

Best regards, Michael.