COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	∂01-Oct-81  1628	Glenn S. Burke <GSB at MIT-ML>
C00009 ENDMK
C⊗;
∂01-Oct-81  1628	Glenn S. Burke <GSB at MIT-ML>
Date: 1 October 1981 19:28-EDT
From: Glenn S. Burke <GSB at MIT-ML>
To: rpg at SU-AI

The files ML:FORMAT;SFMT with second filenames of BRACK, FASL, FLOAT,
NUM, and UMACS should provide you with a translated format.
It appears to work (at least, it is internally consistent).
The autoload things it hacks want to be on the maclisp directory
(referenced by, eg, ((LISP) FORMAT FLOAT) ...)
Note that the canonical place for FORMAT source is on ML, not MC.
For distribution purposes, MC usually (but i don't guarantee it)
has the source for the installed version;  but the differences i believe
are mainly due to Multics hacking.

get format.fas←format;sfmt fasl
get format.bra←format;sfmt brack
get format.flo←format;sfmt float
get format.num←format;sfmt num
get format.uma←format;sfmt umacs

∂03-Oct-81  0024	Grumpy System Building <GSB at MIT-ML> 	Tilde incompatibility in FORMAT for WAITS    
Date: 3 October 1981 03:24-EDT
From: Grumpy System Building <GSB at MIT-ML>
Subject: Tilde incompatibility in FORMAT for WAITS
To: JONL at MIT-MC
cc: BUG-GSB at MIT-ML, rpg at SU-AI

    Date: 2 October 1981 20:47-EDT
    From: Jon L White <JONL at MIT-MC>

        Date: 30 Sep 1981 2206-PDT
        From: Dick Gabriel <RPG at SU-AI>
        Foobar!!! I need to have a format compiled with ascii 26. as the
        tilde character. Is format;format > the source? .  .  .
    This has got to be one of the weirdest incompatibilities I've
    seen in the  past several years.  Of course, the SAIL-specific
    format will have to be  able to interpret code in .FASL files
    compiled elsewhere, so it will still  have to admit ascii 126. as
    another  "tilde" character.
I'm not convinced that this is actually a viable solution.
The problem is one of intent, and goes further than just the "escape"
nature of tilde.  (consider various translations of "~~", "}}", "~}",
and "}~".  Remember also that the LOGICAL sequence "~}" may need to
be recognized specially by format to correctly parse things.)
Anyway, we have an interim version, the production of which i will
eventually automate so as not to need me patching the compilation
environment up.

∂02-Oct-81  2320	Grumpy System Building <GSB at MIT-ML> 	format    
Date: 3 October 1981 02:20-EDT
From: Grumpy System Building <GSB at MIT-ML>
Subject: format
To: rpg at SU-AI

[I don't know/remember what i may have sent you earlier so i'm sending
this now.  Be forwarned that i may not be all here.]
The format i compiled for you was compiled in a relatively crockish
fashion;  as things stood, i was actually using an LSB compiler which
only barely worked, as it was between installations.  (I use, for almost
all of it, common source for all implementations;  recently i've been
working solely on lispm version, so the maclisp has slid.  Needed to
hack it a bit since i need right-curly-bracket, so things were precarious.)
Anyway, with some eventual changes to LSB i expect to make, i expect
that it would not be difficult to simply make a compilation option
which would allow the compilation with the file being read in through
a character-translating stream, and with overriding output file defaults.
(LSB produces multiple output files.  Compiler jcl only accounts for the
fasl file;  i.e., the names of the others are generated automatically,
so compiling while giving the compiler an output filespec can improperly
clobber them.  In theory i can override this.)
Anyway, when i get around to this hack, i'll send you mail.

∂03-Oct-81  1844	Grumpy System Building <GSB at MIT-ML>  
Date: 3 October 1981 21:44-EDT
From: Grumpy System Building <GSB at MIT-ML>
To: rpg at SU-AI

you may try hacking with ml:format;format and the associated files
(FASL, BRACK, FLOAT, NUM, and UMACS).  However ~{...~} will not work
correctly (and note, of course, that you guys can't use stray }s in
your format strings).