New parser branch (Was: Dumping the source DOM?)

Roberto Raggi roberto at
Wed Jul 13 13:06:07 UTC 2005


On Wednesday 13 July 2005 12:41, Vladimir Prus wrote:
> So:
> 1. You do care about things like code refactoring and your parser will
> support that.

> 2. You don't care if you parser can handle Boost. So basically, you're not
> trying to making your parser accept all possible C++ code, but only C++
> subset you're interested in?
it's all about priorities. I'm going to implement the features I need. Boost 
is not a priority for me. Again, PLEASE TRY TO UNDERSTAND that it is 
*impossible* to accept all the possible valid C++ code in KDevelop. 
IDEs work different than compilers. For instance, you can't force a developer 
to write before the "HEADER FILE" than the "SOURCE FILE", than run 
"./configure" generates the makefile(s) because the parser needs the "paths 
and the libraries" than parse "everything" so wait *a lot of time*, and than 
finally popup the code completion list box(.. and do it again for every 

We needs to understand/recover incorrect code. What do you think will be the 
output of gcc-xml if the source code in the KDevelop's texteditor looks like

class X {
   void hello()
        yy.             <<--- your cusrsor is here

a parser for a compiler expects complete code, while an IDE works on 
incomplete code. So we're not interested in well-formed source code. We're 
not writing a compiler.

ciao robe

More information about the KDevelop-devel mailing list