Ian's answer.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
"If you do not pick the simplest explanation, you're being manipulated." —
and if you're being manipulated, you're an agent of EVIL.
For only EVIL specializes in manipulation.
---------- Forwarded message ----------
From: Ian Piumarta <***@gmail.com>
Date: Wed, Aug 12, 2015 at 8:27 AM
Subject: Re: [Fonc] Google Code shutdown
To: Norman Nunley <***@gmail.com>, Faré <***@gmail.com>, Julian
Leviston <***@leviston.net>
Dear Norman, François, Julian,
(Sorry for the direct-mail spam. For silly reasons, I cannot post to
that list any more.)
The code.google repo was always just a copy of my Mercurial repo,
which is (and will continue to be, Network Solutions' occasional
attempts to repossess my domains notwithstanding) reified here:
http://piumarta.com/software/maru/
That page includes both the `original' version (the one on Google
code) and a newer version (which had a better, almost real, compiler
but didn't quite reach the point of self-support before I lost
momentum).
(BTW: there is also a `newest' version, since last year or so, that is
more than 10 times faster OOTB and designed with ease of runtime
dynamic compilation in mind, but it's still not quite finished and
I've been too lazy/procrastinatious [sic] to upload it anywhere.)
I'm not a huge fan of git, or github (or anything that favours
fragmentation over collaboration). There is also no reason to assume
github will be any more permanent than Google code, and lots of
reasons to believe that if/when it goes it will be `lights out without
warning' from which, at least, Google had the good grace (and
PR-sensitivity) to spare us. So I probably should put a link to the
above URL in the Google code page before it freezes over.
Julian: I am very sorry about the lack of comments in my code. When a
brain gets to be as old and tired as mine it only has space left to
remember one way to do any given thing, so looking back at code
written several years ago holds absolutely no surprises; the way I did
anything then is the only way I can imagine doing the same thing now.
I guess code comments are a premature casualty of looming dementia.
;-)
And yes, it would be nice to have a bootstrap from any popular
language. Even better: how about a single abstract general-purpose
`generic' description of how to bootstrap a system, that could
generate a concrete bootstrap in any popular and locally-convenient
language? But that was just the original idea of Maru (or at least of
its immediate predecessor). To date, however, the only `back-ends'
I've managed to make are C and x86 assembler, and neither in a form
that could really be called general-purpose. And the only serious
front end is a textual representation of the intermediate
representation. And the intermediate representation lacks any of the
rigour that such an ideal deserves.
The simplest possible bijections from any GP programming system to
extended lambda calculus, and from ELC to any GP computer hardware,
still eludes.
(FWLIW, I've been trying to do this off and on since 1992 when I
worked for a while on ANDF.
https://en.wikipedia.org/wiki/Architecture_Neutral_Distribution_Format
I still think ANDF's [and TDF/TenDRA before it] big mistake was their
death grip on declarative specification. If they had introduced state
into their model, then made the model executable and self-modifying,
it could have been our missing pair of bijections. Instead dealing
with ANDF felt exactly like trying to write a nontrivial program using
only CPP macros.)
Thanks to all three of you for continuing to show some interest in
this stuff. It should inspire me to take up the cause in some fashion
once more, spare time permitting.
Regards
Ian