Hi Michael,

Michael Albinus <[email protected]> writes:

> Thierry Volpiatto <[email protected]> writes:
>
>> Hi Michael,
>
> Hi Thierry,
>
>> in the last changes you made in tramp, you are raising a user-error
>> message when host have the name of one of the methods.
>> But tramp-dissect-file-name set host to e.g "sudo" when using
>> "/sudo:" which is breaking helm. (It was working fine like this in
>> precedent emacs versions since 2 years now providing also host completion)
>>
>> Is raising an error like this really useful ?
>
> See <http://thread.gmane.org/gmane.emacs.bugs/72217/focus=72255> for the
> story behind.

Hmm, this is another feature that assume everybody (must) use TAB to
have a emacs completion buffer and RET to throw the result when the
minibuffer is completed, which don't help third party tools using
incremental completion like helm.

> In helm, when you complete file names, you shall let-bind `non-essential'
> to t.

Not exactly, I must bind it to t at beginning and let-bind it to nil
when needed in some places.
This is why I never been able to use it up to now.

I will use this instead of defadvicing tramp-dissect-file-name.


BTW I am not a ido user, so I can tell, but double check this error
message is not breaking ido in the same way it broke tramp completion in
helm.
 
Thanks.

-- 
Thierry
Get my Gnupg key:
gpg --keyserver pgp.mit.edu --recv-keys 59F29997 

_______________________________________________
Tramp-devel mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/tramp-devel

Reply via email to