CEDET

Hi,

Misc answer inline below.

On 08/29/2010 01:59 PM, Mandar Mitra wrote:
> Hello,
>
> I'm using cedet-1.0.
>
> I've been trying to set up my .emacs to get the kind of
> completion shown in the images on the CEDET smart completion
> page (http://cedet.sourceforge.net/intellisense.shtml) and Alex
> Ott's Gentle Introduction page.
>
> As suggested, I have included the following lines:
>    (local-set-key "." 'semantic-complete-self-insert)
>    (local-set-key ">" 'semantic-complete-self-insert))
>
> When I use semantic-displayor-traditional, things work fine and on typing .
> or>, I get a *Completions* buffer in which I can choose a completion.
> But if I do the following in custom-set-variables:
>
>   '(semantic-complete-inline-analyzer-displayor-class (quote semantic-displayor-tooltip))

I think this is supposed to be:

(setq semantic-complete-inline-analyzer-displayor-class
       'semantic-displayor-tooltip)

though if you used defcustom you shouldn't put a ' in front to the symbol.

> I get the following message:
> Bug Showing Completions: (wrong-type-argument consp nil)
>
> What I really want is semantic-ia-complete-symbol-menu, but without
> invoking it via an explicit keybinding. Digging around some more I
> found the following snippet
>
> '(semantic-self-insert-show-completion-function (lambda nil (semantic-ia-complete-symbol-menu (point))))
>
> in http://github.com/alexott/emacs-configs/blob/master/rc/emacs-rc-cedet.el.
>
> But this variable doesn't seem to be defined anywhere in the cedet-1.0
> sources. Is this a regression?

I don't remember this variable.  Perhaps Alex has a customization to his 
CEDET somewhere?

> I guess I could mess around with semantic-complete-self-insert, in
> semantic-complete.el but I'd rather not do something adhoc.

The right thing to do is to create a new displayor class for the 
semantic completion engine that uses menus.  If that is used via idle 
completion, I recommend against it as I can imagine the menu being very 
disruptive due to the way it takes over keys.

Creating such a displayor would not be too hard since the tooltip 
version could be adapted to popup a menu by recycling the function 
referenced above.

Eric

Hi,

> I don't remember this variable.  Perhaps Alex has a customization to his 
> CEDET somewhere?

Yes, his reply suggests that this was his own customized version that
apparently didn't work entirely satisfactorily.


> Creating such a displayor would not be too hard since the tooltip 
> version could be adapted to popup a menu by recycling the function 
> referenced above.

I copped out. As suggested by Alex in his reply, I'm going to try out
autocomplete.

You might get others like me though who're going to look at the images 
on the CEDET web-site and try to set things up to work that way. Maybe
it's time to change those pics? Just a suggestion...

Thanks for a great set of tools!
Mandar.

On 08/30/2010 10:58 AM, Mandar Mitra wrote:
>> Creating such a displayor would not be too hard since the tooltip
>> version could be adapted to popup a menu by recycling the function
>> referenced above.
>
> I copped out. As suggested by Alex in his reply, I'm going to try out
> autocomplete.
>
> You might get others like me though who're going to look at the images
> on the CEDET web-site and try to set things up to work that way. Maybe
> it's time to change those pics? Just a suggestion...

Can't they be created with semantic-ia-complete-menu?  Getting that up 
in the other completion engine is a bit different.

I do agree.  We should put the command that makes the pictures on the 
website.

On the whole, I would like to get out of the business of providing fancy 
completion engines, as others have more time to dedicate to the nuances 
of completion.  Getting one of those completion tools that is integrated 
w/ Emacs to be hooked into semantic's idle engine would be good too.

Eric

Eric M. Ludlam wrote (Mon, Aug 30, 2010 at 06:03:27PM -0400):
> >You might get others like me though who're going to look at the images
> >on the CEDET web-site and try to set things up to work that way. Maybe
> >it's time to change those pics? Just a suggestion...
> 
> Can't they be created with semantic-ia-complete-menu?  Getting that
> up in the other completion engine is a bit different.

Yes, indeed they can. If I don't use semantic-complete-self-insert (bound
to . and >), I can type the ., then type a keybinding that brings up
semantic-ia-complete-menu and choose the completion from that. This would
create a situation like those shown in the images.

I was just wondering how I could get . to bring up
semantic-ia-complete-menu, and I couldn't get that to work. Will try some
more and report. 

Mandar.

On 08/31/2010 03:40 AM, Mandar Mitra wrote:
> Eric M. Ludlam wrote (Mon, Aug 30, 2010 at 06:03:27PM -0400):
>>> You might get others like me though who're going to look at the images
>>> on the CEDET web-site and try to set things up to work that way. Maybe
>>> it's time to change those pics? Just a suggestion...
>>
>> Can't they be created with semantic-ia-complete-menu?  Getting that
>> up in the other completion engine is a bit different.
>
> Yes, indeed they can. If I don't use semantic-complete-self-insert (bound
> to . and>), I can type the ., then type a keybinding that brings up
> semantic-ia-complete-menu and choose the completion from that. This would
> create a situation like those shown in the images.
>
> I was just wondering how I could get . to bring up
> semantic-ia-complete-menu, and I couldn't get that to work. Will try some
> more and report.

Ah, that should be straightforward.  The easiest would be to copy 
semantic-complete-self-insert and replace the call to 
semantic-complete-analyze-inline with semantic-ia-complete-menu.  Maybe 
rename the fcn to something else too.

If this is considered generally useful, it can be added to semantic-ia.el.

Eric

Eric M. Ludlam wrote (Tue, Aug 31, 2010 at 07:13:01PM -0400):

> Ah, that should be straightforward.  The easiest would be to copy
> semantic-complete-self-insert and replace the call to
> semantic-complete-analyze-inline with semantic-ia-complete-menu.
> Maybe rename the fcn to something else too.

Many thanks for the tip. I did this, ran make, and things work the way I
wanted. I've attached a patch in case it's any use (I've changed the name
and docstring a little).


> If this is considered generally useful, it can be added to semantic-ia.el.

I made the change in semantic-complete.el, since that's where the original
semantic-complete-self-insert was. I'm a little uncomfortable about having
two different functions that share almost all their code, but don't know
enough elisp to clean this up. Ideally, I suppose the user should be able
to use customize in order to choose between semantic-complete-analyze-inline 
and semantic-ia-complete-menu. 

Thanks again,
Mandar.