Re: [anjuta-devel] RFC: Symbol-db future

[Johannes Schmid <jhs jsschmid de>]

Hi!

> we would need some real cases where the method works and where not.
> You mentioned <gtk.h>. In that case, stopping the scanning at the first 
> level, you won't get anything useful to completion.
> We may set a trade-off of two or three levels at most. Would it be 
> sufficient?

Actually I was thinking about infinite levels (=> scan all files
included) because that is exactly what "gcc -MM filename" gives us. I
think any limitation in level would make things not work for some
people.

You can play a bit with the gcc command if you like, you need to give
the correct CFLAGS though (but we have the packages so we should be able
to do that). For the cases I tried this can be up to 200 files which we
need to query in the database (and rescan if they aren't already in the
database though I think after the user opened 5 files we should have
everything).

For me the main question is if we can do a query on the global database
(containing all header files we ever visited) that only gives us the
symbols of the files we want and how fast this query is. If it isn't
fast enough we might think of creating a temporary database that will
speed up subsequent queries.

IMHO doing the search based on packages would be a nice step forward but
I feel that we might want to drop the intermediate think and do the real
solution now.

Just want to make clear that I want to hear everybody's opinion and that
I would like to do a decision together based on the best arguments.

Thanks and regards,
Johannes