Re: RFC 196 (v1) More direct syntax for hashes

On Thu, 7 Sep 2000, Michael G Schwern wrote:

> On Wed, Sep 06, 2000 at 06:40:10PM -0000, Perl6 RFC Librarian wrote:
> > C<scalar(%hash)> should return what C<scalar(keys %hash)> currently
> > returns.
> 
> Good.

	I'll weigh in with a "me too" on this.


> > C<reset %hash> should reset the hash iterator, instead of calling
> > C<keys> or C<values> as is currently the case.
> 
> Sounds good, except the name.  reset() already does something.
> Currently, reset() is for clearing large swaths of global variables (a
> dubious feature) and for reseting ?pattern? searches.  Two weird
> features crammed into one keyword.  Shoving a third in doesn't seem
> pleasent.

	I think my suggestion would be that we remove the other two
features (which I doubt anyone uses all that often).  However, if lots of
people chime in with "no, no!  We *must* have our reset as-is!" (or, more
likely, if removing these features is too much of a pain for the p52p6
people), then here are some suggestions for alternate keywords:

	renew, restart, refresh, freshen, 


> > The parser should special-case the variations of C<sort %hash> so
> > that it returns the keys and value, calling the comparison function
> > for keys.
> 
> I'm trying to think of a common case where I'd want sort() to treat a
> %hash like a list (as it currently does) but I can't really think of
> one.

	I would go so far as to argue that the current behaviour is
actually _bad_...if I want a list, I'll use a list.  If I put things into
a %hash, that means that I want THESE keys to be associated with THOSE
values.  The way sort currently works destroys those linkages.


> What you proposed and what you put in your example doesn't quite play
> out.  C<sort keys %hash> returns just the sorted keys of %hash, but
> your foreach example implies that both keys and values would be
> returned, sorted by key.  Both would be useful.  
> 
> I'd edge towards the latter because the former is simple enough to
> reproduce (just explicitly say C<sort keys %hash>) and its use is
> mostly when you want to iterate over the keys of a sorted hash, which
> would largely be eliminated if your proposal returned a 'sorted' hash.


	Again, "me too".


				Dave