Sean O'Rourke sorourke@cs.ucsd.edu writes:
Luke Gorrie lgorrie@common-lisp.net writes:
- (slime-autodoc-mode): New command.
I think we're losing here by rolling our own instead of using eldoc. For example, today I noticed that slime-autodoc continues to echo during an isearch; eldoc has been made to avoid doing this.
Sorry, slip-of-the-finger plus Gnus' send-without-confirm behavior led me to send the incomplete message above. To continue...
In slime-autodoc's favor, it _is_ much simpler, particularly when you consider having to use an advice to take advantage of eldoc in pre-21.3 Emacs. But I suspect that there will be other cases of unforseen gotchas and minor behavioral differences in the future, and can't help but think that maintaining a parallel implementation of eldoc is more trouble than necessary.
/s
On Sun, 30 Nov 2003, Sean O'Rourke wrote:
Sean O'Rourke sorourke@cs.ucsd.edu writes:
Luke Gorrie lgorrie@common-lisp.net writes:
- (slime-autodoc-mode): New command.
I think we're losing here by rolling our own instead of using eldoc. For example, today I noticed that slime-autodoc continues to echo during an isearch; eldoc has been made to avoid doing this.
Yes, it's certainly a bit too eager; but checking
(and (not executing-kbd-macro) (not cursor-in-echo-area) (not (eq (selected-window) (minibuffer-window))) ...)
would probably be sufficient to overcome this. Though I also wouldn't mind a timer to delay showing the arglist for half a sec or so. (And, while we are at it, packaging autodoc as a minor mode that can be turned on/off in inferior-lisp and slime-repl too, wouldn't hurt either.) But all these can be solved quite easily without eldoc.
In slime-autodoc's favor, it _is_ much simpler, particularly when you consider having to use an advice to take advantage of eldoc in pre-21.3 Emacs. But I suspect that there will be other cases of unforseen gotchas and minor behavioral differences in the future, and can't help but think that maintaining a parallel implementation of eldoc is more trouble than necessary.
But maybe it's better to just wait and see. (And perhaps by the time it turns out that eldoc is indispensable, everybody will use Emacs v. >= 21.3 :-))
Andras
Andras Simon andras@renyi.hu writes:
Yes, it's certainly a bit too eager; but checking
(and (not executing-kbd-macro) (not cursor-in-echo-area) (not (eq (selected-window) (minibuffer-window))) ...)
Thanks, committed.
The real problem is that I skimped off on doing a proper replacement for eldoc, I still think it only requires a tiny amount of code and is a better option than advising.
I've turned it on in my .emacs so I'll make it more usable soon.
Cheers, Luke
Andras Simon andras@renyi.hu writes:
Though I also wouldn't mind a timer to delay showing the arglist for half a sec or so. (And, while we are at it, packaging autodoc as a minor mode that can be turned on/off in inferior-lisp and slime-repl too, wouldn't hurt either.) But all these can be solved quite easily without eldoc.
Good idea. That's now in as `slime-autodoc-delay' with a default of 0.2 seconds. This avoids doing lots of RPCs to Lisp when you're scrolling around, much better.
Just have to put back support for looking up variable documentation and it should be respectable I thik.
-Luke
-
Andras Simon -
Luke Gorrie -
Sean O'Rourke