Re: [Gnu-arch-users] Re: Tla spork

[Tom Lord]

Throughought these threads, few people have discussed the following
topics and the quality of the discussion has suffered as a result:

* the distinction between "s-exps as abstract data type"
  and "s-exps as syntax" 

* what kinds of abstract data types to use for arch configuration
  and similar data

* whether and how parsers and printers are isolated or isolatable
  in arch;  whether they can be customized per-user in some cases;
  the costs in difficulties of playing with parsers and printers

* what *external* design critera to apply to configuration data design
  (e.g., discussion of what else must parse and print this data is not
  absent but i haven't seen in depth discussion of how the data can be
  usefully representable in other environments).

There is also (still!, sign) a lot of sloppiness confusing what I've
been discussing with lisp.

>From the perspective of someone largely though not entirely sitting
this thread out:  it's the quality of the discussion, not the quality
of the furth proposals, that is the primary bug.   

This is probably mostly my fault, actually, for two reasons:

With the series of furth-related posts I was looking pretty far ahead.
There's a few first steps which are far more minor purturbations to
arch --- I wanted to think far enough ahead to be personally confident
that those steps were in a plausibly good direction.

My first (arguable) mistake was to invite the list in on some that
thinking.  I got next to 0 useful feedback from it (but with two huge
exceptions, so this was arguably also not a mistake).  It mostly just
alarmed people, but not in a way that generated any kind of thoughtful
analysis.

Instead, and this relates to my second mistake, people were largely
alarmed reflexively by an apparent "language war".  I'm sure I
provoked that feeling of alarm (unintentially) because of what might
be deemed "lisper's oppression": lisper engineers are *so* accustomed
to being told their full of shit (only to then go on and prove that
their approaches are right, if they can get past the shit) that I
think when I started to present the general idea of furth, something I
see as little more than a general direction in the design space of
tiny evaluative interpreters, I implicitly sent the message "Ok,
presumably a bunch of people will give me shit for no good reason
now...."    And surprise!  I got what I expected!

BTW: as a result of the unfortunate reactions my posts generated, i've
(at least for now) mostly suspended giving this list updates about
my longer term plans, focussing mostly on the tiny steps being turned
into code.   For example, it's remarkable that people are still
griping about furth since, although it yielded many fruitful thoughts
that turned into xl, furth itself doesn't figure largely in my
arch-specific plans.   

In addition to all of the other bugs in this thread: y'all are tilting
at windmills.

-t