source browsing and more

[Thomas Schilling]

> Hi Victor,
>
> On Friday 15 February 2002 16:12, Victor Röder wrote:
> >
> > Hey, very cool! I'm curios about the persistant class store (or is it
not
> > part of your wrapper?).
> > My examinations are partically completed and I will have some time for
> > Gideon again.
> >
>
> There is some time until the wrapper gets a gideon plugin, but it is
> basically a KDE interface to sourcenav functionality.
>
> sourcenav maintains a symbol database, which is generated using a fuzzy
> parser so that you can browse the source code realtime, ie as you are
> writing. I can for instance directly go to the include hierarchy, or the
call
> function graph, or the class hierarchy and move in the code like that.
> sourcenav has a project file, and lots of database files in a directory
> called .snprj. Very similar to what sniff++ does ;)
>
> The great thing about sourcenav is that it can completely replace the
current
> class-store. My intention is to make it optional... like the editor
> plugins... We'll see how that works, I don't want to destroy anybody
else's
> code. Though I don't know how we could allow multiple class-store
> implementations to be used, maybe another bridge pattern here too. We can
> separate the interface and the implementation.
>
> Anyway, I'm going to make the interface components quite modular, so that
> they can be used in any code.
>
> Thanks,

How much does this support code completion? I.e. it would be best if
_every_ scope (_all_ code blocks in "{ ... }"  also while-scopes, if -
scopes)
is put into the store (including local typdefs, vars, ...) including its
position
of declaration and definition. That way the code completion parser could
easily find out the current scope.