Andreas Raab wrote:
>>How about the really, REALLY obvious one?  The units used by TeX?
>>Used successfuly on everything from 16-bit 640k DOS to supercomputers
>>and useful for a wide range of output devices, yet fitting within a
>>reasonable range of integers.
> 
> 
> Okay, that's probably TSQOTD[*] but what exactly _are_ the units used by TeX
> internally? I at least have no idea whatsoever as I've never had the need to
> think about how TeX represents my real world units internally ;-)
> 
> Cheers,
>   - Andreas
> 
> [*] TSQOTD: The Stupid Question Of The Day

Not stupid at all - I never bothered to find out before. Here is what I 
found: "Internally in TeX, dimensions are stored 32-bit values, 
normalized so that 1 pt corresponds to the scaled point (sp) value of 
65536." And elsewhere:
	"pt	Point, 1/72.27
	bp	Big pt., = 1/72
	sp 	Scaled point = 1/65536 pt"

So the largest magnitude would be 23 meters. This sounds large enough 
for paper size and maximizes accuracy given 32 bits. And who would we be 
to argue with DK? Except maybe for the general superiority of the metric 
system. There are good arguments for it even in the land of typography: 	
	http://www.cl.cam.ac.uk/~mgk25/metric-typo/

Anyway, I still stand by my 1 micrometer = 1 unit proposal, because we 
aren't really going to need more precision - we are just handing these 
numbers to the OS's print driver, we don't want to do heavy math on them.

-- Bert