Trash mechanism

[Dave Cridland]

On Wed Aug 25 15:33:22 2004, Rahul Sundaram wrote:
> --- Dave Cridland <dave at cridland.net> wrote:
> > On Wed Aug 25 13:57:07 2004, Rahul Sundaram wrote:
> > > For non trivial standards its better to use the
> > same
> > > implementation if possible. It would be a good way
> > to
> > > avoid misinterpretrations and implementation
> > glitches
> > > I totally disagree.
> > > If the specification is widely misinterpreted, then
> > the specification > is broken.
> 
> Let me explain. Lets take the freedesktop menu spec as
> an example. its a relatively big spec and everytime a
> implementation is created by kde,gnome,xfce and rox
> there is a chance that they may do something slightly
> different where things might be ambigous or where they
> interpt it differently

Yeah, so where this happens, the specification needs clarifying.

I'm not sure this *has* happened as yet, though.


> > > If a particular implementation has glitches or
> > misinterpretations, > the implementation is broken.
> > 
> sure. the implementation can be considered broken. I
> am just saying that the we have a high chance of
> having some obscure bugs in the implementation. whats
> wrong with sharing code as much as possible to avoid
> these problems
> 
> 
Sharing the code is fine. It's a choice any implementation can make, 
and in many cases it's a better idea. Indeed, it pretty much always 
happens - there'll be one implementation for *every* GNOME 
application, for instance. Cool, eh?

But the important thing to get right is the specification - whether 
code happens to be shared or not isn't something that should concern 
us. I'd personally distrust a specification which had only been 
implemented once unless it was utterly trivial.

Dave.