c++ code completion status report

Eray Ozkural (exa) erayo at cs.bilkent.edu.tr
Mon Jan 7 10:36:02 UTC 2002

Hash: SHA1

On Monday 07 January 2002 08:09, Loy wrote:
> > Is that 0.5 seconds going to grow when we are
> > dealing with complex codes,
> > such as one that #include's a good collection of kde
> > libraries?
> >
> > Fearsome.
> another approach, aside from timed refresh, is by
> monitoring the TIMESTAMP of these files.  rarely would
> a system screw-up on its timestamp of files, so its
> reliable.

Well yes, I'd mentioned that. You have to do a dependency analysis on each 
processed file and store the output of type analysis. The difficult issue 
being here is that type analysis is _not_ a tree of class information. I 
still maintain that you have to write 1/2 of the complete C++ semantic 
analysis for proper code completion. Whether it's worth it, I have no idea.

OTOH, many other IDEs that do code completion would seem to rely on compiler 
specific features. I recall that Borland's code completion was indeed 
complete whereas Microsoft's code completion had occasional success. I 
suggest we refrain from using a weak method wherever possible.

I'm hearing that Thomas is writing a C++ grammar. That is doable, but the 
hard part is the semantics. Is he constructing complete C++ type information 
as well?


- -- 
Eray Ozkural (exa) <erayo at cs.bilkent.edu.tr>
Comp. Sci. Dept., Bilkent University, Ankara
www: http://www.cs.bilkent.edu.tr/~erayo
GPG public key fingerprint: 360C 852F 88B0 A745 F31B  EA0F 7C07 AE16 874D 539C
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org


More information about the KDevelop-devel mailing list