RFC: Structure of background parser and language support

[Kris Wong]

> As I've said on irc, we don't want ICore to be global static because
> it's an interface and I think language controller is a good owner for
> the bg parser (and maybe duchain as well) as bg parser is certainly
> language-oriented.

As it seems we don't want to use singleton objects, I will change the background
parser implementation.  I think there are a few others that didn't get that memo,
though, as it's all over the language support and also in the editor integrator code.
ICore is not a singleton object.  ICore simply has a static pointer that is set in the
constructor and cleard in the desctructor.  It was to allow access to ICore from the
backbground parser without having to set it somehow.  This may no longer be 
necessary once I change it.
 
Kris Wong
-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 4049 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20070717/d2ddc71d/attachment.bin>