Re: [cedet-semantic] CEDET problems

[Eric M. L. <eri...@gm...>]

On 08/04/2011 03:15 AM, Roberto E. Vargas Caballero wrote:
>
>> I don't think there is anything wrong.  What happens is that cscope and
>> global can only be used to identify in a symbol you look for *exists*.
>> Once the symbol is known to exist, that faux tag is returned.
>>
>> If the tool you use does something like they symref output buffer, then
>> each file needs to be pinged to load in the line where the match occurs.
>>    The side effect is the regular parsing activities of semantic.
>>
>> If the tool you use is a completion engine or the like, then once the
>> tag is known to exist, the limited information provided by global or
>> cscope must be augmented with type information and other details, which
>> also requires parsing the source file.
>>
>> The *good* news is that you are parsing many fewer files than without
>> those databases available.
>
>
>
> But you can see how in cscope and globals tables there is 0 tags in the
> databases, so I think this databases doesn't help anything (or I lost something
> here?)

The Global and CScope tables always have 0 tags stored in Emacs.  I can 
certainly see how that would be confusing.  I'll look into replacing the 
'tags' extra text with something else so it doesn't suggest a problem.

, and I don't understand why ebrowse databases doesn't have information
> about usual functions (at least semantic can know where is defined and parse the
> file where label is defined) and only have information about class and structs.

The ebrowse parser (the one between ebrowse and Semantic) has some 
issues.  I don't remember what they all are anymore.  It works ok for 
some basic cases, but I know it gets problematic with more complex stuff.

Eric