Path: ...!news.mixmin.net!aioe.org!eternal-september.org!feeder.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: anton@mips.complang.tuwien.ac.at (Anton Ertl)
Newsgroups: comp.lang.forth
Subject: CREATE...DOES> etc. and flash (was: Dictionary separated ...)
Date: Thu, 28 Jul 2016 09:17:00 GMT
Organization: Institut fuer Computersprachen, Technische Universitaet Wien
Lines: 82
Message-ID: <2016Jul28.111700@mips.complang.tuwien.ac.at>
References: <3d8511c7-8cbf-4ef5-bdc3-eb33c5399d42@googlegroups.com> <8c63b476-e4fb-491e-a69a-873b51e3d02a@googlegroups.com> <nn1bq0$1m7u$1@gioia.aioe.org> <e5334450-3055-4dcc-9b27-06568c9f855e@googlegroups.com> <nn4hac$1i03$1@gioia.aioe.org> <04KdnWxKbtCzUgjKnZ2dnUU7-RPNnZ2d@supernews.com> <nnabiq$4qh$1@gioia.aioe.org> <LY2dnf9BXvXOmQTKnZ2dnUU7-XHNnZ2d@supernews.com>
Injection-Info: mx02.eternal-september.org; posting-host="d47d3421039fe8026514328ad0ebacae";
	logging-data="21204"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/yU5p1F+trXAZZmhz6qGuo"
X-newsreader: xrn 10.00-beta-3
Cancel-Lock: sha1:hPOXtcswa0yCwCrS622qqYMvGS8=
Bytes: 4277

"Elizabeth D. Rather" <erather@forth.com> writes:
>On 7/27/16 3:08 AM, HAA wrote:
>> Elizabeth D. Rather wrote:
>>> Historical note: ;: was Chuck's original name for the structure which
>>> for many years now has been CREATE DOES> (by separating the creation
>>> from the instance behaviors you get more flexibility).
....
>I don't really remember the details, but as I recall it didn't have the 
>separate CREATE step,

That's interesting, because I have been thinking about how to deal
with the problem that Forth systems compiling to flash have
difficulties with CREATE...DOES>, because DOES> overwrites a field
that ideally would lie in flash (and which we don't want to
overwrite).

With the existence of quotations, a way to do this would be to replace

: foo
  create code1
does>
  code2 ;

with

: foo
  [: code2 ;] <does-create code1 ;

With this syntax there would be no intermediate stage where the
created word just pushes an address, and whichj has later to be
changed into a word that runs code2.

However, that's not the only issue in this context; we also have other
words that change existing words, in particular, IMMEDIATE; among
non-standard words, we have e.g., COMPILE-ONLY and SET-OPTIMIZER/OPT:.

Doing all of that in one step is not very practical.  We could have a
word with one parameter for each of these options that creates a word
with these options in one step, maybe something like:

new-word ( flag1 flag2 xt1 xt2 xt3 xt4 "name" -- )
\ flag1 specifies immediacy
\ flag2 specifies compile-only
\ xt1 specifies interpretation/execution semantics
\ xt2 specifies compilation semantics
\ xt3 specifies what to do when COMPILE,ing the xt of "name" (for optimization)
\ xt4 specifies TO "name" semantics

But, apart from the sheer uglyness, that would require standardizing
all these features, and leaving no room for non-standard features.  So
I lean more towards recording the things for the next definition in a
RAM buffer, and then the next defining word puts all these features in
the newly-defined word (in flash, in one step), and wipe the (RAM)
slate clean for the next word to be defined; i.e. for an immediate
word, it would be something like:

next-immediate : bar ... ;

For the DOES>-using foo example above it would be:

: foo
  [: code2 ;] next-does create code1 ;

For an optimizing 2DUP it would be

:noname drop postpone over postpone over ; next-optimizer
: 2dup over over ;

And then one could combine several of these setup words.  E.g., for
defining <http://forth-standard.org/standard/facility/PlusFIELD>:

: +field ( n1 n2 "name" -- n3 )
  [: @ + ;] next-does
  [: >body @ ?dup if postpone literal postpone + then ;] next-optimizer
  create over , + ;

- anton
-- 
M. Anton Ertl  http://www.complang.tuwien.ac.at/anton/home.html
comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html
     New standard: http://www.forth200x.org/forth200x.html
   EuroForth 2016: http://www.euroforth.org/ef16/