[Reduce-algebra-developers] Changes wrt the character "#"

[Arthur N. <ac...@ca...>]

I have just checked in a bunch of updates that alter the exact treatment 
of "#".

Firstly the preprocessor-like directives #if, #else, #elif, #endif, #eval 
and #define may now be written without a preceeding exclamation mark...
but note that they are NOT recognised within quoted items (even if those
items spread across several lines), so
     '!#if
still needs the "!" and does not try to trigger conditional inclusion,
and
     a := '(
       one
     #if nil
       two
     #endif
       three);
does not have the #if and #endif processed (because the quote mark 
protects them).

Secondly # can be used to notate characters that might otherwise be hard 
to express. So far the follow-through that uses these is not in place, but 
I hope to address that sometime!
     #name;
where name is one of the entity names from HTML5 will stand for one of teh 
about 2000 symbols available there. RIGHT NOW you might only use things 
that stand for characters with codes up to 255, and if you try larger 
values they can crask the Lisp system maybe. So eg "#amp;" for "&" is OK.
Sometime later I hope that this will extend to #alpha; etc.
     #hhh; or #xhhh;
where hhh are hex digits (and the x can be in upper or lower case) will
be for a Unicode code-point specified in hexadecimal.
     #uddd; or #Uddd;
will be a Unicode code point specified is decimal.

There are a few HTML5 names made up just of letters a-f which could look 
like hex specifications. The name takes precedence and you can use X or a 
leading 0 to forcibly indicate numeric input. If the number you specify is 
over 0x0010ffff then what you gave could not be valud Unicode and it will 
be rejected. Unrecognised names, overlarge numbers, missing semicolons and 
the like should just cause the characters to go through to Reduce 
unaltered.

A way you can try this today (and this WILL CHANGE soon I hope) is to just 
go
    lisp "#xc2;#xa3;";
which gives the two byte utf-8 sequence for a a pounds-sterling sign, and 
if you have a utf-8 enabled terminal you should see that.

There are a few places where this change is not 100% upwards compatible 
(and I needed to alter a few source files scattered through Reduce to 
survive). Mainly a sequence of characters "#d;" where d is any string of 
hex digits suddently means something special. I found two places where 
this arise naturally.
In susy2 there was a line
     operator !#a,!#b,!#c; 
where the final #c; hurt. I could fix that by just putting an apparently
frivolous blank before the semicolon.
In the factoriser some names like !#a were used with the intent that they 
would never clash with normal sensible names. By altering them to use !#z 
I could remove the clash with hex-specfied characters.
Any code that uses "#" as an infix operator could find itself writing
     a#b;
but if instead it speced things out as
     a # b;
it will be safe.

              Arthur