[texhax] (no subject)

[Robin Fairbairns]

> At 11:13 AM 6/7/2003 -0400, Stewart C. Russell wrote:
> >Herbert Gintis wrote:
> >>... the language he wrote it in is a complete mess. ...  the
> >>absence of real floating point

knuth, to first order, doesn't write much tex _code_.  so i presume
you're talking about the implementation language.  knuth wrote tex in
pascal, which _does_ have provision for floating point.  i'm retiring
hurt on this assertion.

> >you don't want floating point in typesetting software. The tiniest 
> >precision difference can make a line break differently, which can make a 
> >page throw differently, which can wreck the pagination of a whole book.

carelessly used, this can indeed happen.  i doubt somehow that, had
don chosen to use floating point, it would have shown serious signs of
carelessness.  however, one of the reasons tex has ported so well and
easily is that it hasn't had to be concerned about the various
floating point formats (and fpu bugs) that have shown their heads
since the mid-70s.

>          Different users want different things. TeX lacks graphics support. 

cf. sunder on c.t.t, whose thesis (roughly) seems to be that the whole
exercise is a failure because it can't do graphics.

> When I wrote graphics routines for TeX, I had to import a floating point 
> package, and it worked, though it's incredibly clunky to use.
>          Try drawing curves, circles, etc. without floating point!

done it.  firmware of graphics devices...

rational arithmetic, c'était moi!

>          The power should be in the software, and the user can decide 
> whether to take advantage of it, I think.

when i wrote commercial graphics software for a living, the most
intractable bugs were those around the error bars of my floating point
calculations and the interface to the graphical database.

i can see why knuth didn't do it, but i agree, it's a pain sometimes.
i don't even know whether the omega-ists are providing it.