Signaling text-read-only

Signaling text-read-only

[Deniz Dogan]

I have a situation here where I need to tell the user that the part of
the buffer where point is located is read-only. This is what I am
thinking right now:

(unless (rcirc-looking-at-input)
  (error (error-message-string (list 'text-read-only)))

or:

(unless (rcirc-looking-at-input)
  (signal 'text-read-only nil)))

I haven't found any examples of this explicit sort of signaling in the
other Lisp files in Emacs, but I might have missed something.

The reason why I need this is that `rcirc-complete' shouldn't be
called in read-only areas of the buffer.

What is the right way to do this?

-- 
Deniz Dogan

Re: Signaling text-read-only

[Leo]

On 2011-02-15 11:35 +0800, Deniz Dogan wrote:
> I have a situation here where I need to tell the user that the part of
> the buffer where point is located is read-only. This is what I am
> thinking right now:
>
> (unless (rcirc-looking-at-input)
>   (error (error-message-string (list 'text-read-only)))
>
> or:
>
> (unless (rcirc-looking-at-input)
>   (signal 'text-read-only nil)))
>
> I haven't found any examples of this explicit sort of signaling in the
> other Lisp files in Emacs, but I might have missed something.
>
> The reason why I need this is that `rcirc-complete' shouldn't be
> called in read-only areas of the buffer.
>
> What is the right way to do this?

The question arose when Deniz and I discussed the patch here:
http://paste.pocoo.org/raw/gUNWgGTy4zMSI0ApR3vZ/ in #emacs.

As required by completion-at-point-functions, it makes sense for
rcirc-completion-at-point to return nil when re-search-backward errs and
therefore for rcirc-complete to handle the case when completion table is
nil so I think the patch is applicable regardless of whether or not to
signal an error.

I think it is good idea to give some feedback to user but I think
text-read-only is not the right one (the error is not caused by
text-read-only). It should probably be something like:

(unless (rcirc-looking-at-input)
  (error "Point not located after rcirc prompt"))

to be added at the very beginning of rcirc-complete function body.

What do you think?

Leo

Re: Signaling text-read-only

[Deniz Dogan]

2011/2/15 Leo <sdl.web@gmail.com>:
> On 2011-02-15 11:35 +0800, Deniz Dogan wrote:
>> I have a situation here where I need to tell the user that the part of
>> the buffer where point is located is read-only. This is what I am
>> thinking right now:
>>
>> (unless (rcirc-looking-at-input)
>>   (error (error-message-string (list 'text-read-only)))
>>
>> or:
>>
>> (unless (rcirc-looking-at-input)
>>   (signal 'text-read-only nil)))
>>
>> I haven't found any examples of this explicit sort of signaling in the
>> other Lisp files in Emacs, but I might have missed something.
>>
>> The reason why I need this is that `rcirc-complete' shouldn't be
>> called in read-only areas of the buffer.
>>
>> What is the right way to do this?
>
> The question arose when Deniz and I discussed the patch here:
> http://paste.pocoo.org/raw/gUNWgGTy4zMSI0ApR3vZ/ in #emacs.
>
> As required by completion-at-point-functions, it makes sense for
> rcirc-completion-at-point to return nil when re-search-backward errs and
> therefore for rcirc-complete to handle the case when completion table is
> nil so I think the patch is applicable regardless of whether or not to
> signal an error.
>
> I think it is good idea to give some feedback to user but I think
> text-read-only is not the right one (the error is not caused by
> text-read-only). It should probably be something like:
>
> (unless (rcirc-looking-at-input)
>  (error "Point not located after rcirc prompt"))
>
> to be added at the very beginning of rcirc-complete function body.
>
> What do you think?
>

I'm fine with your error suggestion as well. I just figured it would
make sense to signal text-read-only, since whatever the command
completed to wouldn't be inserted at point, as that part of the buffer
is read-only.

-- 
Deniz Dogan

Re: Signaling text-read-only

[Leo]

On 2011-02-15 18:38 +0800, Deniz Dogan wrote:
> I'm fine with your error suggestion as well. I just figured it would
> make sense to signal text-read-only, since whatever the command
> completed to wouldn't be inserted at point, as that part of the buffer
> is read-only.

I installed something as discussed. Feel free to tweak the error message
if you think 'text-read-only is better.

Cheers,
Leo