[patch] Re: kconfiggroup template 'mess'

[Oswald Buddenhagen]

On Fri, Oct 26, 2007 at 10:30:08AM +0200, Thiago Macieira wrote:
> Em Thursday 25 October 2007 21:29:41 Oswald Buddenhagen escreveu:
> > On Thu, Oct 25, 2007 at 09:21:31PM +0200, Jaroslaw Staniek wrote:
> > > Oswald Buddenhagen said the following, On 2007-10-25 19:24:
> > >> On Thu, Oct 25, 2007 at 05:06:22PM +0200, Jaroslaw Staniek wrote:
> > >>> Can we start with this and polish it a bit?
> > >>
> > >> i'm not impressed. major ugliness as a tribute to redmond.
> > >
> > > As Thiago and then I have mentioned - template/nontemplate
> > > versions of function (here: readEntry/writeEntry) cannot be mixed.
> >
> > you just reiterated the same statement which is have already
> > demonstrated wrong in this broadness.
> >
> I didn't say that the overloading of templates with non-templates is
> the problem. I even know for a fact that older Microsoft compilers
> (VC6) can have them too

> (provided you put the template argument somewhere in the argument
> list).
> 
a-ha! now we are getting somewhere. :)

> But I did recommend splitting the template vs non-template names, as a
> way of cleaning up the code and trying to find out why the Microsoft
> compilers were getting internal errors.
> 
that's sure a valid technique to identify the problem.

> It was a wild guess, but if it makes the code more readable and
> compiles, I'm all for it.
> 
that's the problem - it doesn't. it virtually kills all the elegance
templates have introduced in the first place.

on a related note ... somebody wants to check how our templace code
performs wrt code size (duplicated instantiations)? qt4 supposedly fixed
big issues present in qt3 and i'd prefer not to repeat the mistakes done
before.

-- 
Hi! I'm a .signature virus! Copy me into your ~/.signature, please!
--
Chaos, panic, and disorder - my work here is done.