Re: [CEDET-devel] completion in semantic

[David P. <dav...@wa...>]

Eric,

[...]
 > I'll check these changes in.

Thanks!  All is working fine now :-)

IMO, semantic-complete is close to be able to replace senator
completion ;-)

A feature it would be nice to have in semantic-complete, is
`senator-completion-uniquify-names', very useful to disambiguate tag
names.

A short example in a Java buffer:

Using `semantic-complete-test' on "get" symbol returns the following
completion buffer:

Possible completions are:
getMethod0                         getMethod1
getMethod2                         getMissingParameters
getURL                             getValue
getText                            getValue
getValue

You can see that "getValue" is proposed three times, and it is
quasi impossible to know which one is the good choice :-(

In a such case (for example, using `senator-jump' on "get" symbol)
senator completion proposes this:

Possible completions are:
getMethod0                         getMethod1
getMethod2                         getMissingParameters
getText                            getURL
getValue().Essai                   getValue().Essai2
getValue().Internal2

It is now easy to distinguish each occurrence of "getValue", and to
choose the right one.

Maybe would it be possible to reuse the
`senator-completion-uniquify-names' code in semantic-complete?
Unfortunately, I don't know well enough (yet) your new code to be
able to do that myself.  Maybe could you give me a starting point,
and I could try something?

David