Grant Rettke scripsit:

> I had kind of
> envisioned R7RS wg2 as determined entirely by a big table of "major"
> (by some definition) implementations of R5RS and R6RS where you look
> at a feature or SRFI, look who provides it, see that it is obvious for
> wg2, and maybe follow it by some or no discussion, and then go along
> with the standardization process.

I've always tried to avoid defining "major" implementations.  If you look
at ImplementationContrasts, I started out by counting Schemes that did
or didn't do something in particular, but dropped that pretty quickly in
favor of just listing what all the Schemes I could get access to actually
do, and letting other people draw conclusions about their importance in
the Scheme of things.

I did use a process generally similar to this to develop a big list of
packages that I thought might be included in WG2.  (But I also looked at
Java and Python, not wishing to exclude something useful just because
Schemes hadn't done it yet.)  I then asked the original WG2 members to
vote on that list; the results wound up on ConsentDocket, StandardDocket,
and RejectedDocket.  After that, I worked on proposals (the various pages
ending in Cowan on the wiki), but eventually gave up because I couldn't
get feedback, and asked the SC to hibernate WG2 until WG1 was done and
more energy would be freed up.  In the meantime, though, when I got a
new idea I put it on InputDocket, and just a month ago transferred most
of those to StandardDocket in batches, using unanimous consent.

> I kind of look at it starting from the end; the goal of wg2 is to
> produce a usable batteries included system, so for people who are not
> intimately involved with Scheme, what would they want/need to know to
> see why it is that way.

I agree, but I don't think the "why" is necessarily accounted for by
what existing Schemes provide.  I want the large language to go beyond
what Schemes have provided, to include not only batteries but the
entire Erector/Meccano set.

> Part of it strikes me as kind of an odd fit though, as though it is
> slightly different than the goal of wg2, one of agreement or shared
> desire. WG2 seems more like, report on what is the current state of
> things.

As noted above, I want to go beyond the Scheme state of the art.

> FFI is kind of an easy one to pick on; the discussion always converges
> on "it is too hard to standardize" and yet I bet that the majority of
> distributions provide it, so that is kind of a curiosity, is it
> obvious whether it should be included, or not? From an srfi
> perspective it seems a bad fit, but from practical wg2 goals, it seems
> perfect.

I think the original WG thought it was too big a rathole to go down,
and requires too much discussion of fundamentals.  For one thing, there
is a big split between FFIs that allow access to arbitrary C (or Java
or C#) routines using Scheme glue, and those that allow access only to
C or Java or C# glue which must be written to invoke arbitrary routines.
There are also ugly issues around callbacks, GC, and call/cc.

> Is the SRFI process meant more for implementers, or users, or both?

Users, clearly.  It's a Request For Implementation process.  However,
the person calling for implementation must in most cases provide
at least *an* implementation.

> Who is WG2 serving, users, implementers, or both?

Users first and foremost.  Of course, implementers matter too, because
a standard is only useful if it's implemented.

> Thanks for your patience bearing with me learning about how this is all going.

Sure.  I'll probably boil this thread down into a wiki page at some point.

--
John Cowan     http://ccil.org/~cowan    cowan@ccil.org
Monday we watch-a Firefly's house, but he no come out.  He wasn't home.
Tuesday we go to the ball game, but he fool us.  He no show up.  Wednesday he
go to the ball game, and we fool him.  We no show up.  Thursday was a
double-header.  Nobody show up.  Friday it rained all day.  There was no ball
game, so we stayed home and we listened to it on-a the radio.  --Chicolini

_______________________________________________
Scheme-reports mailing list
Scheme-reports@scheme-reports.org
http://lists.scheme-reports.org/cgi-bin/mailman/listinfo/scheme-reports