> What does "blessing with official status" mean? It may be considered
> official by those submitting libraries to the repo, but it may still be
> ignored by other parts of the Scheme community.

It's about advertising and trust, basically. Nothing in Scheme is "de
jure" official except the RnRS reports, and that shouldn't change.

The SRFI process is "de facto" official (i.e. universally known and
widely respected) but as you say, only a few individual SRFI documents
are official (either de facto, i.e. widely implemented, or de jure, i.e.
become part of an RnRS standard).

> This already happens with SRFIs. In the end, a SRFI only reflect the
> opinion of its author (which may have been influenced by other opinions)
> but this doesn't mean that there is any consensus about implementing
> that SRFI.

SRFI approaches things from the implementations' point of view. A
fast-moving library collection approaches things from the opposite point
of view, that of the users. Users who want some code can immediately
write it and add it to the collection. If there are problems with it, as
there probably are, those can then be fixed until stability is reached.

The end result would be the same: stable and widely implemented
libraries. But we'd arrive at the result from the opposite direction.

> The only thing where consensus could be reached among the various
> implementations (R6RS, R7RS, Racket, Guile, etc.) are library
> namespaces, I think.

Good point. Namespaces are important.

> So publishing a SRFI at least claims a global namespace of the form
> (srfi nnn). The same is true for the R7RS-large process, which claims
> namespaces of the form (scheme xxx). Then we have (rnrs xxx) for
> descendants of R6RS.
>
> We could add another universal namespace for your idea, Lassi.

Indeed, there should be one.

But the main problem is to have only one such namespace :)

I could write my own library collection, but I'm not motivated to do
that if it just becomes one more collection added to the pile of the
existing collections. I would be quite motivated to contribute to a
collection that is trusted as de facto official. I have been motivated
to write SRFIs because SRFI enjoys such respect, but as a process I
think the evidence we have from the last couple of years shows that most
problems would benefit from starting with a lighter process.

So basically, combine the reputation/trust and community of SRFI + the
development style of Schemepunk, Spells, Yuni, etc.