Intellisense(tm) again

holle at almaden.ibm.com holle at almaden.ibm.com
Tue Jun 22 19:39:37 BST 1999



Hmm, good point. Maybe we should only parse the files included by this file ?
This means running cpp on our current file and the grepping for the class
QString declaration ...  Emacs only looks at all it's buffers when he tries to
complete things.
---- You can even preparse the includefiles once the user wrote "#include
<foo.h>" (or that string is loaded from disk).
Furthermore you could actually do some ai-algorithm here that matches QString to
the headerfile qstring.h and looks in that headerfile/related database first

-holger


Herwin Jan Steehouwer <herwinjs at palet.nl> on 22/06/99 10:44:16

Please respond to kdevelop at fara3.cs.uni-potsdam.de

To:   kdevelop at fara3.cs.uni-potsdam.de
cc:    (bcc: Holger Lehmann/Almaden/Contr/IBM)
Subject:  Intellisense(tm) again





Hi,

I have been walking trough the source for that Intellisense(IS) stuff..

i have found some struff:

1) When i add a variable in a functions:

void Class::test()
{
     QString bla;
     bla.

When i what that IS see that bla is a QString, it has to scan this file
first to know what type bla is.. maby faster to make in the parser a
function to parse only 1 function !

2) we need to have ALL the KDE and QT functions in a database, what we can
do i a create a file on first startup that generate that file/database.
If you want to upgrade we can make a rescan option..

I think we have only to scan the .h files of QT and KDE !


some food for thought ;-)


HJ



Herwin Jan Steehouwer
  herwinjs at palet.nl                 http://www.caiw.nl/~herwinjs
  steehouwer at kde.org              KDE developer http://www.kde.org







More information about the KDevelop mailing list