From:	 adrian@cox.rr.com (Adrian Mariano)
To:	 mjhammel@lwn.net
Subject: Re: On the Desktop... 'units' (3 Jul 01 LWN) (fwd)
Date:	 Thu, 5 Jul 2001 19:21:44 -0400

Thank you for your comment on GNU units.

The conversion shown is correct.  A change in 79 degrees Fahrenheit is
equal to a change of 26 degrees Celsius.

Quoting the very first paragraph of the manual:

| OVERVIEW OF `UNITS'
|        The `units' program converts quantities expressed in vari-
|        ous scales to their  equivalents  in  other  scales.   The
|        `units'  program  can  only  handle  multiplicative  scale
|        changes.   For  example,  it  cannot  convert  Celsius  to
|        Fahrenheit  but  it  can  convert  temperature differences
|        between those temperature scales.

GNU units version 1.55 is the most recent stable release.  Version
1.77 is the current beta testing release.  It does support conversion
between temperature scales.  Beta testing releases are appearing at
http://www.cam.cornell.edu/~adrian/units.html. I did not know that
anybody had put any of them on the GNU ftp site.

Version 1.77 includes a completely and radically rewritten parser so
there are lots of opportunities for new bugs.  Supporting temperature
conversions (rather than temperature difference conversions) was very
difficult to do in a natural way.  People blithely assume that
absolute temperatures are converted and write things like

> Apparently this is due to the units.dat database, which seems to omit the
> '32' in the usual formulas:
>   (degF - 32) * (5/9) = degC
>   (degC * (9/5)) + 32 = degF

I ask these people: What should be the result of the conversion
     17 joules degF^3 / kg m 
to
     calories kelvin degC degF / lb ft
and how can it be determined from those formula above?