[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] Reformulated numeric-tower ballot

On Fri, May 16, 2014 at 10:32:01AM +0200, Sascha Ziemann wrote:
> 2014-05-15 12:41 GMT+02:00 Alaric Snell-Pym <alaric@x>:
> >
> > I suspect that a "common consensus" on the set of SRFIs that makes a
> > "practical implementation" will emerge organically, and predicting that
> > set to mandate it now will be difficult :-)
> >
> This is the reason why the decision which SRFIs are important and which
> are not important is useless. Instead it would be much more important to
> specify some kind of Maven like package repository for Scheme. There
> have been several attempts, but Racket has its own, Chicken has its own,
> Gamit has its own. And if you try to switch from one Scheme to another,
> it is always a pain to find the packages necessary for the program. Not
> having access to package repositories is nowadays a major productivity
> drawback. If R7RS wants to be large/great, it should do something great.
> And joining all Scheme code would be the greatest thing R7RS could do.

Agreed.  Just declaring a grab-bag of libraries as "standardized" isn't
going to work, especially considering the current tendency is towards
marginalizing many Scheme implementations by doing things like declaring
Unicode and the full numeric tower as mandatory.  Instead, it should aim
towards unification.

I also agree with Wingo's remarks: this standard is just a bunch of
folks who like to quibble about language esoterics, instead of
implementors who know their stuff, and have a feeling for the way
standards can help or hinder implementation and experimentation.
The scope is also way too large, which means people like me won't have
the energy to keep up with all proposals; by the end, the only people
left will be those with nothing better to do than argue about standards.
This is turning out to be truly "design by committee".  I'm also going
to stop participating in this standardization effort.  If it creates
anything useful that matches my needs for a library, I might try to
make a compliant implementation that but otherwise this seems like a
complete waste of effort.


Scheme-reports mailing list