Re: [CEDET-devel] semantic-analizer : new C++ joy!

[Bruce S. <bru...@ce...>]

"Eric M. Ludlam" <er...@si...> writes:

[...]

> The basic problem is that the ebrowse DB is set up as a "system"
> database.  Thus, if you had:
>
> #include <ec.h>
>
> it would likely work.

Ah!  I'd forgotten semantic-default-c-path, I think.  Does that make
sense?  I think I'm beginning to get it to work, anyway.

[...]

>>(I note it's easy enough to get apparently duplicated databases.  I
>>seem to have files
>>
>>    !home!bruce-test!test!h!BROWSE
>>    !home!bruce-test!test!h!BROWSE-load.el
>>    ~!test!hBROWSE
>>    ~!test!hBROWSE-load.el
>>
>>Shouldn't there be some canonicalisation?)
>
> Huh.  I haven't seen that before.  The ~ should never show up in the
> file names there.  How'd you do that?

Just (semanticdb-create-ebrowse-database "~/test/h").  I see there's
lots of use of expand-file-name in semanticdb-file.el, so I guess this
ebrowse functionality is cunningly avoiding it, somehow.

Hmm, it doesn't look as though semanticdb-file-name-directory makes
its argument absolute.  (Or that it's even supposed to (the docstring
suggests it returns a relative directory, though if
semanticdb-default-save-directory is non-nil the returned value will
be an absolute pathname).)

[...]