Re: Proposal for 2.8: Glog

[Matthias Clasen <mclasen redhat com>]

On Wed, 2005-05-04 at 15:14 +0100, Tim M�wrote:
> On Wednesday 04 May 2005 14:15, Matthias Clasen wrote:
> 
> > If log categories are valuable, how about adding simple category support
> > to g_log, instead of adding a clone of the (IMO) somewhat overengineered
> > log4x stuff ?
> 
> GLog isn't really a log4* clone. It does logging and it does have categories, 
> yes, but that doesn't make it a clone yet. GLog does not have appenders, it 
> does not have configurable layouters, nor does it use a config file. Also, 
> categories in GLog do not have hierarchical relationships, they are just 
> simple names.
> 

Sounds much better then. 
I did not really look much deeper than "has categories".

> Looking at log4c, I feel GLog gets the balance pretty much right. It's simple 
> and straight forward and more powerful than g_log(), but not overengineered 
> (with 500 lines of code plus convenience macros there isn't too much danger 
> of that).
> 
> More importantly, GLog was designed with the GObject type system in mind. One 
> might even want to extend that a little bit and provide an additional hook to 
> register custom per-GType print functions or something like that.
> 
> I am not sure how it is possible to extend the existing g_log() system in a 
> backwards-compatible way without ending up with something pretty close to 
> GLog, but maybe that's just my lack of creativity :-)

One could add another family of g_log functions taking an extra category
parameter...