Joseph Wright writes:
> On 19/04/2011 22:36, Frank Mittelbach wrote:
> > Bruno Le Floch writes:
> > > > Within xtemplate, you'll find some places where we use the fact that the
> > > > keys do have catcodes to pull off some 'special effects'. Now, we also
> > > > remove spaces there, so spaces can also be used for said 'special
> > > > effects'. But such a change would require some alterations.
> > >
> > > Ok, so my assumption was wrong. But really, playing with catcodes is evil
> > > ;-).
> > right and that was a fairly recent change in implementation, wasn't it. It
> > might be worth while rethinking if we limit the nature of keys and their
> > comparison even if we have to modify the xtemplate code for it
> I'd have to check the detail, but from memory this was the most
> convenient way to store the key order within the template itself. There
> is nothing to stop us moving it back out again to a dedicated location.
I'm not claiming that the change to a restricted key set as suggested is the
"right thing" to do, but I agree with Bruno's sentiment on the evility :-) of
catcodes and thereis something to say for restricting "keys" to be string-like
objects, ie behaving like they "look".