[GCC-XML]New cv-qualified type referencing scheme

[BUCHMANN Richard]

I never seen information about the QualifiedLocation node before. I am
missing something ?

Anyway, I suggest to forget the backward compatibility in the main GCC_XML
output.
We could write a XSL file to do the compatibility job. The old GCC_XML user
need to have a XSL tools.

The main advantage of this approach is that the future maintenance should be
increased significantly.
The disadvantage is that the old user need to install a new tools and add 2
lines in his makefile to convert the new format to the old one.

Richard Buchmann

----- Original Message -----
From: "Brad King" <brad.king at kitware.com>
To: "Tony Cook" <tony at bci.gb.com>
Cc: "GCC-XML Discussion" <gccxml at gccxml.org>
Sent: Friday, October 17, 2003 4:22 PM
Subject: Re: [GCC-XML]New cv-qualified type referencing scheme


> On Fri, 17 Oct 2003, Tony Cook wrote:
>
> > Could the same be done with the location attribute (e.g.
> > location="f0:305"). That too violates the ID/IDREF spec when linking to
> > <File id="f0" ... /> though this might cause an explosion of
> > intermediate tags if you use your approach!
> >
> > <QualifiedLocation id="f0:305" file="f0" line="305" />
> >
> > Lots of <QualifiedLocation ... /> tags will appear. Would an alternative
> > solution be to add two additional attributes to all elements supporting
> > the location attribute. E.g.
>
> Fixing thd ID/IDREF problem for this case is probably a good idea. I agree
> that this would cause an explosion of elements.  It would effectively
> double the total number of element dumped because almost every declaration
> has its own line number.  This is not acceptable.
>
> > <...... location="f0:305" file="f0" line="305"/>
> >
> > So here 'file' is now the nominated IDREF. In my case this addition
> > would not affect my XSLT stylesheets, though it might affect other
> > parsers that barf on unexpected attributes ... thoughts?
>
> There is one function that dumps the location attribute.  It should be
> pretty easy to change it to dump what you suggest.  Unlike the type
> change, this change will duplicate information in order to maintain
> backward compatibility.  It is probably an acceptable compromise, though.
> I'll implement this change when I get a chance unless someone suggests a
> better approach.
>
> Thanks,
> -Brad
>
> _______________________________________________
> gccxml mailing list
> gccxml at gccxml.org
> http://www.gccxml.org/mailman/listinfo/gccxml
>