Re: FRAC

[Ralf Hemmecke]

> I'm probably not being clear. The implementation of FRAC is in FRAC.spad.
> So we have the following:
> 1) FRAC.spad: Contains the real genuine definition of FRAC.
> 2) DIVRING.ap: Contains what would be in init_FRAC.spad, except that there is a
> dependency cycle DIVRING -> ? -> FRAC -> ? -> DIVRING
> 3) FRAC.ap: Declaration 'extend FRAC(xxx): yyy'. Note there is no
> need for this to be a definition (ie. extend FRAC(xxx): yyy == zzz').
> 4) axextend.as: May extend fraction (can't remember).
>
> When axiom runs, it should only look for function implementations in
> (1) and (4) above. The other files are just there to compile
> libaxiom.

Aha! That means that libaxiom.al does *not* contain all of the axiom
algebra. I didn't know that.

I thought, that all the steps that are done to get libaxiom.al basically
translate the whole content of the axiom library into .ap files (which
are basically .as files, just in another format, but (up to the
documentation part) equivalent), right?

Now with the above information, it becomes more clear.

If later

)co aaa.as

is called then the aldor compiler must have knowledge of what functions
there are in the algebra library. Only that knowledge is in libaxiom.al,
but (mostly) not the functions themselves, right?

Let's suppose aaa.as defines a new domain AAA. That will be available to
the interpreter, but will it be also accessible to bbb.spad for

)co bbb.spad

? How? With a )library command in bbb.spad?

And the other way round? Could I use some domain BBB from bbb.spad in
aaa.as and compile it for Axiom?

At least I doubt the latter is possible.

Ralf