Man this is a trap. There is no logical relation between failing to
provide examples and the the validaty of a feature.
The consideration about maintenance makes more sense. and I would like
to say again, I agree it is a large patch and I understand any repulsion
toward it: after all it is alien and foreign.
A big patch solving one or two case is useless.
You've already spend so much commits patching all the mess with
[NoArrayLength] [array_length_pos] [cname] and [FloatingReference].
There is really no point to introduce another mess to replace them and
to kill their descendants. A movement in the opposite direction costs
too much. Let's forget about it.
Regards,
Yu
On Fri, 2008-08-22 at 12:26 +0200, Jürg Billeter wrote:
> On Fri, 2008-08-22 at 05:49 -0400, Yu Feng wrote:
> > Just a scratch of idea:
> >
> > class array<T> {
> > [CCode (snippet = "int i; while(((T)this)[i]) i++; return i;")]
> > public int get_length();
> This won't work (arrays may contain null/0 entries), and it would be
> slow.
>
> > [CCode (snippet = "gpointer * rt; while((T)(this)[i]) i++; result.length
> > = i; rt = g_new0(T, i); for(i=0; i<result.length; i++) rt[i] = this[i]);
> > return rt;")]
> > public array duplicate();
>
> This would at least break memory management, and it looks like a strange
> combination of C and Vala syntax, not easy to understand.
> > [CCode (snippet = "return this[id];")]
> > public T element(int id);
>
> Might work if we treated the element method specially, doesn't sound
> that straight forward to me.
>
> > [CCode (snippet = "g_free(this);")]
> > public void free();
>
> We don't want free functions in the bindings API, we already have
> `delete' to free memory if you use raw pointers.
>
> Jürg
>
_______________________________________________
Vala-list mailing list
Vala-list@gnome.org
http://mail.gnome.org/mailman/listinfo/vala-list
