commit 8147d3c27c breaks grep over tramp

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

commit 8147d3c27c breaks grep over tramp

John Shahid
Hi all,

Looks like commit 8147d3c27c breaks `grep' when ran on a remote host
using tramp.  Previous to the commit I would see the list of grep hits,
but after this commit there is a command prompt as part of the grep
hits.  This makes the first hit unclickable.  I bisected the history to
find the offending commit and found the following change to be the
cause:

> @@ -2912,8 +2903,7 @@ tramp-sh-handle-make-process
>      ;; otherwise we might be interrupted by
>      ;; `verify-visited-file-modtime'.
>      (let ((buffer-undo-list t)
> -  (inhibit-read-only t)
> -  (mark (point-max)))
> +  (inhibit-read-only t))
>        (clear-visited-file-modtime)
>        (narrow-to-region (point-max) (point-max))
>        ;; We call `tramp-maybe-open-connection', in
> @@ -2926,9 +2916,7 @@ tramp-sh-handle-make-process
>   (let ((pid (tramp-send-command-and-read v "echo $$")))
>    (process-put p 'remote-pid pid)
>    (tramp-set-connection-property p "remote-pid" pid))
> - (widen)
> - (delete-region mark (point-max))
> - (narrow-to-region (point-max) (point-max))
> + (delete-region (point-min) (point-max))

Is anyone else seeing the same thing or is it something with my
configuration ?

Cheers,

JS

Reply | Threaded
Open this post in threaded view
|

Re: commit 8147d3c27c breaks grep over tramp

Michael Albinus
John Shahid <[hidden email]> writes:

> Hi all,

Hi John,

> Looks like commit 8147d3c27c breaks `grep' when ran on a remote host
> using tramp.  Previous to the commit I would see the list of grep hits,
> but after this commit there is a command prompt as part of the grep
> hits.  This makes the first hit unclickable.  I bisected the history to
> find the offending commit and found the following change to be the
> cause:

Will check. My respective commit message was

    (tramp-sh-handle-make-process): Simplify.

Maybe this part of the patch could simply be withdrawn. I'm short in
time these days, so I cannot promise it will happen before the weekend,
but who knows ...

> Is anyone else seeing the same thing or is it something with my
> configuration ?

I do :-(

Obviously, tramp-tests.el must be extended, because they've passed
before I've committed the above patch.

> Cheers,
>
> JS

Best regards, Michael.

Reply | Threaded
Open this post in threaded view
|

Re: commit 8147d3c27c breaks grep over tramp

John Shahid


[...]

> Maybe this part of the patch could simply be withdrawn. I'm short in
> time these days, so I cannot promise it will happen before the weekend,
> but who knows ...

Thanks for getting back to me.  Take your time, I will revert this
change locally or use a previous commit for now.

>
>> Is anyone else seeing the same thing or is it something with my
>> configuration ?
>
> I do :-(
>
> Obviously, tramp-tests.el must be extended, because they've passed
> before I've committed the above patch.

I will take a look to see if I can add the missing test coverage.

Cheers,

JS

Reply | Threaded
Open this post in threaded view
|

Re: commit 8147d3c27c breaks grep over tramp

Michael Albinus
John Shahid <[hidden email]> writes:

Hi John,

>> Maybe this part of the patch could simply be withdrawn. I'm short in
>> time these days, so I cannot promise it will happen before the weekend,
>> but who knows ...
>
> Thanks for getting back to me.  Take your time, I will revert this
> change locally or use a previous commit for now.

Should be fixed now in master. I've reverted the change in
tramp-sh-handle-make-process, and I've added a comment to stop me (and
other people) doing the same error, again.

> Cheers,
>
> JS

Best regards, Michael.

Reply | Threaded
Open this post in threaded view
|

Re: commit 8147d3c27c breaks grep over tramp

John Shahid

Michael Albinus <[hidden email]> writes:

[...]

> Should be fixed now in master. I've reverted the change in
> tramp-sh-handle-make-process, and I've added a comment to stop me (and
> other people) doing the same error, again.

Thank you.

JS