defontifying program

[RWK@FUJI.ILA.Dialnet.Symbolics.COM (Robert W. Kerns)]

    Date: Fri, 3 Aug 90 16:18 EDT
    From: miller@gem.cam.nist.gov (Bruce R. Miller)
    The 1st occurrence an  `turns on' the font machinery and defines a font
    association. Such as:
       D,#TD1PsT[Begin using 006 escapes](1 0 (NIL 0) (:SWISS :ROMAN :SMALLER) "HL8") 

Actually, it's not the first occurrance of an epsilon.  Rather,
it's the first occurrance of this entire string:
D,#TD1PsT[Begin using 006 escapes]
This greatly reduces the odds of accidentally mistaking
line noise or greek letters or something for the start
of fonted text.  If the entire string doesn't match, then
it's not enabled.

(I'm the one who designed the format).

Actually, it's just a bit hairier than that:  The 
in the string is the version number of the protocol
(that's the *ONLY* version in use).  The part after
the  is not part of the "start fonts", but is checked
(and an error signaled if it doesn't match, I believe).
It's purpose is to avoid people looking at the files
thinking that the filesystem had trashed the file somehow.
(Which happened a couple times when the protocol was first
introduced, until I added that text string!)

This was designed under a set of constraints relating to
forward and backward compatibility, the need to store fonted
text on foreign systems, the need to be able to write unfonted
text on foreign systems without any strangenesses, etc.

    which turns on fonts and sets font # 1 to be swiss.roman.smaller.  
    The next use of font # 1 would just use 1, as before.
    0 switches back to the `default font'.

    The first use of font # 2 might be something like
	 (2 0 (NIL 0) (:SWISS :ITALIC NIL) "HL12I")
    And so on...

Those are examples of font index definitions.  Font index
references can be just (1) or (2), or for indexes 0 to
9, can be just 1 or 2.

Also, don't forget to turn  into just a single .  You
can't just always remove the first character after a !

There's also a way (, I believe) to turn OFF the mechanism.
This is used when appending text files, when you don't know
if there's epsilon encoding already going on, or what the
indexes are mapped to.  Since you don't know what the status
is, first it must turn it *ON*, in case it wasn't on before,
and then it can turn it off.

    BTW: Somebody posted a request for us to include Signatures in our mail
    messages (but neglected to include a return address :>).  I'm all for
    that, but does anybody know how to get such things Automagically stuck
    in?  I mean, you dont really expect me to THINK everytime I mail a
    message, do you????

I absolutely refuse to add those silly signature lines to the end
of the messages, *AND I WISH EVERYONE ELSE WOULD STOP!*.  They waste
space & time and aren't cute or funny any more.  Computer mail is more
than 20 years old now, and buying, stealing, or writing brain-dead software
just doesn't cut it as an excuse any more!  I can sympathize with one
stuck in the dilemma, but I think it has to be so rare as to not justify
imposing such a burden on hundreds of people every time they send one of
their hundreds of message.

On the other hand, I usually *DO* refrain from using Symbolics text
styles in mail messages to the Slug list, because I'm well aware that
many Slug readers (not just a few) don't read their mail on a Lispm.