XXXX wrote:
> how flexible is the 'standard' units file that comes with frink? I'd
> like to add other units (mostly to do with energy densities, masses
> of materials, etc).

The standard units file is quite flexible. Hopefully, you shouldn't
have to modify the units in there (unless you're, say, a surveyor and
you want the "foot" to mean the "surveyfoot" .)

It's rather easy to make your own list of additional definitions that
will complement the standard distribution. You can then load this file
using the

-k [filename]

switch from the command line, or using the "use file" menu option
from the interactive GUI.

I'd be happy to add or distribute libraries that you might build.

> so.. what are the limits on the standard unit set?

There are very few limits. You can even add new base dimensions at
runtime. So, if you're working on warp drive physics, and you need to
add "cochranes" as a measure of subspace distortion, you can do it.

On the other hand, Frink runs on a variety of handheld devices, and
one of the things that contributes most to its startup time and memory
usage is the size of the standard data file, so I have to consider what
units get added somewhat carefully.

> And is there a
> possibility of having standard distribution 'libraries' which contain
> more detailed units for a given subject (ie: physics, chemistry, etc)
> that you could require on demand?

Absolutely. I recently added the ability to have libraries
distributed within the standard distribution. At the moment, there
aren't any, but I'll add them as they get good enough (and if they're
not too big.) For the moment, I just have them downloadable from the
web site.

Some of the big questions are:

1.) Do these get written in an object-oriented fashion? For example,
if we want to store the properties of Carbon, it might have a density, a
number of protons and neutrons, etc., all of which may vary. (They may
vary by isotope, or by temperature and pressure, or by allotropic form
such as graphite or diamonds or buckyballs. Getting properties of these
objects might have to be represented by functions.) Do we do this as
new units such as "uraniumdensity" or as properties of an object like
Uranium which is an instance of ... uh... what class? What should the
inheritance hierarchy look like, if there is one?

Since I haven't fully implemented inheritance, there's necessarily
some delay in doing that.

2.) How stable and good do the libraries have to be to be allowed into
the standard distribution?

3.) Should we have an external repository of libraries and a tool to
download these files like Perl's CPAN or yum or apt-get? Should these
be downloaded on a user-by-user basis or for the whole machine? Should
they be downloaded on first request?

Sometimes I look at the whole CRC Handbook of Chemistry and Physics
and despair at how much effort it'll take to get all of that into Frink,
but hopefully it'll happen someday.

If someone has libraries to contribute, or would like some help in
designing them, please let me know!