Re: how to implement function copy-subword-to-irc-buffer

["B. T. Raven" <nihil@nihilo.net>]

Kevin Rodgers wrote:
> B. T. Raven wrote:
>> Kevin Rodgers wrote:
>>> B. T. Raven wrote:
>>>> I think I need (thing-at-point 'word), (buffer-list), set point and
>>>> mark, copy-region-as-kill-no-mark, and other usual suspects to
>>>> implement
>>>> this but really want to automatically copy the substring of the word
>>>> before point or the whole word if point is in whitespace after the
>>>> word.
>>>> This (sub)string should go to the point in the irc buffer (one whose
>>>> buffer-name starts with #) and then make that irc buffer current.
>>>
>>> Looking at rcirc.el, it does not look like that is a reliable method to
>>> select
>>> "the" IRC buffer.  It actually supports multiple IRC buffers,
>>> maintained in
>>> rcirc-buffer-alist (which is local to the server buffer, of which
>>> there may
>>> also be more than 1).
>>>
>>> Glossing over that for the moment, how about:
>>
>> I was glossing over that too. Just to get started I was planning on
>> using the function with only one open server and channel. In fact, I
>> probably will never need it to go to more than one irc buffer. It's just
>> to move foreign language dictionary entries quickly into a particular
>> channel during a live session.
> 
> The code below simply selects the first buffer returned by buffer-list
> whose
> name starts with "#" and has an rcirc server buffer associated with it.
> 
>>> (let* ((subword (if (looking-at "\\>")
>>>             (thing-at-point 'word)
>>
>> ;; already the \> regex was new to me. If I C-s to the word "the" RET,
>> ;; then point is already past the word
> 
> Yes, just as you requested: point is after the word.
> 
>>>           (buffer-substring (or (car (bounds-of-thing-at-point 'word))
>>>                     (point))
>>>                     (point))))
>>
>> I can't get my mind around this either; why the (or ... ? Don't I always
>> want car of the cons returned by bounds-of-thing-at-point ?
> 
> Yes, but there are corner cases where it may be nil, which causes
> buffer-substring to signal an error.
> 
>>>        (buffer-list (buffer-list))
>>>        (irc-buffer (catch 'irc-buffer
>>>              (while buffer-list
>>>                (when (and (string-match "^#"
>>>                         (buffer-name (car buffer-list)))
>>>                   (buffer-local-value 'rcirc-server-buffer
>>>                               (car buffer-list)))
>>>              (throw 'irc-buffer (car buffer-list)))
>>>                (setq buffer-list (cdr buffer-list))))))
>>>   (when irc-buffer
>>>     (funcall rcirc-switch-to-buffer-function irc-buffer)
>>>     (insert subword)))
>>
>> I couldn't wrap this in a defun and get it to work because I am in ver.
>> 22.3 Can I copy rcirc.el from /emacs23/lisp/net over to ver 22.3?  If
>> not, is it safe to just overwrite all the ver 22.3 files, directories,
>> etc. with the equivalent ver. 23 ones. I tried to use my old .emacs with
>> 23.1 but it wouldn't work because of (ucs-tables) in my .emacs.
>>
>> Anyway, thanks a lot for the code. I may have to postpone any more study
>> of it until I dare to move everything to ver. 23.
> 
> No it is not safe.  If you want to experiment with such an approach, put
> the newer files in the site-lisp directory.

Good idea. I will temporarily rename rcirc.el and .elc in .../lisp/net
and put ver. 23 copy in emacs22/site-lisp.

> 
> But as far as I know, there are no dependencies in that code on version
> 23 -- in fact, there couldn't be, as I am running 22.3 as well and
> tested it successfully.
> 
> You should be able to wrap that in a defun and put it in your emacs as is.
> What happened when you tried?

After connecting to server and channel I see:

and: Symbol's value as variable is void: rcirc-server-buffer

server buffer name is irc.dal.net:6667 in buffer list but
Atlanta.ga.us.dal.net in mode line. I suppose that mismatch is normal?
Nothing was copied out of dictionary buffer into channel buffer at point
(ERC>)


> 
>>>> Is this doable? I want to assign this function to some keychord but am
>>>> not sure what would be suitable (in keeping with traditional key
>>>> assignment philosophy). I have standard binding for almost every thing
>>>> but single character cursor movement. Is there a way of getting a list
>>>> of all key combos (and/or ranges of key combos) that are not currently
>>>> bound?
>>> From the Keymaps node of the Emacs manual:
>>>
>>> ,----
>>> |    As a user, you can redefine any key; but it is usually best to
>>> stick
>>> | to key sequences that consist of `C-c' followed by a letter (upper or
>>> | lower case).  These keys are "reserved for users," so they won't
>>> | conflict with any properly designed Emacs extension.  The function
>>> keys
>>> | <F5> through <F9> are also reserved for users.  If you redefine some
>>> | other key, your definition may be overridden by certain extensions or
>>> | major modes which redefine the same key.
>>> |
>>> `----
>>>
>>
>> It may be overridden but from what I know now it probably won't be. For
>> instance  if I type (at random) C-h k and then C-M S-h I find that the
>> shifted combo has been translated to C-M h but C-h S-k (C-h K)isn't
>> translated to C-h k. If I avoid C-M combos and known prefixes (C-x, C-h,
>> and ESC) then most combos will be undefined (M-s, C-s, shifted versions
>> of these, and Alt combos if there are keyboards with both Alt and Meta).
>> Emacs didn't use to discriminate between shifted and unshifted combos
>> but now it seems to in many contexts.
>>
>> I thought there might be a way to get the complement to the set of
>> defined key bindings. Maybe there is no easy way to do that.
> 
> If you want to use a binding that is "in keeping with traditional key
> assignment philosophy", then use C-c LETTER.

Yeah, I know but I'm already using all of those. But those bindingS are
just for inserting single unicode characters so they could be reassigned
now that ver. 23 represents unicode internally.
>