Re: Indexer (was Re: [cdt-dev] CDT 3.0 new features)

[John Camelon <jcamelon@xxxxxxxxxx>]

Leon,

We are well aware of several shortcomings
with the current architecture and technology choices, and we are looking
forward to providing a competitive solution in CDT 3.0.  
To aid us in this endeavour, Bogdan
and I would appreciate a bug report with a concrete code example (non-proprietary
of course) so that we can solve your problem without too much more guesswork.
 

Thanks,
JohnC
www.eclipse.org/cdt

cdt-dev-admin@xxxxxxxxxxx wrote on 12/01/2004 03:55:38
AM:

> On Tuesday 30 November 2004 18:33, John Camelon wrote:
> John,
>    I read carefully the 49783 bug content. I know that unprofessional
advices 
> are terrible, therefore I do not advice, but just remind that, for
example, 
> the technology used in Source Navigator is very efficient and works
fast on 
> relatively large projects. My last test case was RTEMS source tree
which has 
> about 12000 files in 120MB of source.
> 
> Thanks!
> 
> P.S. Just wanted to clarify for myself: the case when function "body"
resides 
> in another file (C or C++) which is the *.cpp/*.c file is the case
Iuse very 
> often. Is this something unusual?
> 
> > Leon,
> >
> > Open Declaration (F3) does not use the index.   This decision
was done at
> > the time to improve the performance of the feature for C++. Unfortunately,
> > it costs us correctness in C code which is structured as you
have.
> >
> > The file is parsed to figure out exactly what ASTNode we are
looking at.
> > If the file is included, we are able to navigate to the node
quite easily.
> >  However, if the parser does not see the file, we are unfortunately
out of
> > luck without a larger context.
> >
> > Our 3.0 plan includes the implementation of a new AST which (almost)
> > already allows for "fuzzy" bindings in C.
> > Selection/Navigation features will be moved to this AST at some
point, and
> > I'll ping you at that time to try it out.
> >
> > Cheers,
> > JohnC
> > www.eclipse.org/cdt
> >
> > cdt-dev-admin@xxxxxxxxxxx wrote on 11/30/2004 11:24:52 AM:
> > > On Tuesday 30 November 2004 18:06, Bogdan Gheorghe wrote:
> > > > > More specific. For example, C code with functions
been called from
> >
> > one
> >
> > > > > file, while function declaration is in another
- always causes "the
> > > > > operation is unavailable on the current selection".
> > > >
> > > > 2 things come to mind:
> > > >         1. Do you specifically
include the header file you are trying
> >
> > to
> >
> > > > F3 to in the source file?
> > >
> > >    These are both *.c files inside the same managed
project. One
> >
> > contains the
> >
> > > functions calls, another - functions bodies. There is also
one header
> >
> > file
> >
> > > (included in both *.c files) which contains the functions
prototypes.
> > >
> > > >         2. If the source and header
files are located in different
> >
> > folders
> >
> > > > in your project, are your include paths set up?
> > >
> > >    Everything is inside the same directory/folder.
> > >
> > > > Our current framework doesn't handle cases where files
are not
> >
> > included
> >
> > > > directly as of now. This is on the to do list for 3.0
(Bug 49783).
> > >
> > > Seems to be my case, if I understand you correctly, no?
> > > But even if so, I should receive the function prototype
in header
> > > file in this
> > > case, while I receive nothing.
> > >
> > > Thanks.
> > > --
> > > Leon M.Pollak
> > > leonp at plris dot com
> > >
> > > _______________________________________________
> > > cdt-dev mailing list
> > > cdt-dev@xxxxxxxxxxx
> > > http://dev.eclipse.org/mailman/listinfo/cdt-dev
> 
> -- 
> Leon M.Pollak
> leonp at plris dot com
> 
> _______________________________________________
> cdt-dev mailing list
> cdt-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/cdt-dev