Laurent (and Puiseux series without explicit variable specification

[Ralf Hemmecke]

I would like to create a domain that behaves (approximately) like
UnivariateLaurentSeries, but does not need specify the variable and the
center. So I created

----
)abbrev domain QETALS QEtaLaurentSeries
QEtaLaurentSeries(C: Ring): Exports == Implementation where
Exports ==> UnivariateLaurentSeriesCategory C
Implementation ==> UnivariateLaurentSeries(C, 'q, 0$C)
----

The compilation seems to succeed but yields an irritating

--------------non extending category----------------------
.. QEtaLaurentSeries(#1) of cat
(|UnivariateLaurentSeriesCategory| |#1|) has no
(|UnivariateLaurentSeriesConstructorCategory|
|#1|
(|UnivariateTaylorSeries| |#1|
|q|
(|Sel| |#1| (|Zero|))))

(1) What does the compiler try to tell me?
(2) Do I have to worry that the compiled code does not do what
I expect it to do?

In fact, of course eventually, I will need access to the underlying
taylorRep, but instead of relying on

taylorRep: % -> UTS

I will implement my own access function and probably just "pretend" that
the representation is appropriate, because I explicitly program it that way.

So the goal is to have univariate Taylor/Laurent/Puiseux series
constructors that do not need that just work like the ones that are
already in FriCAS, but where the variable is implicitly set to 'q and
the center to 0.
(Puiseux series are a "nice to have", but no "must".)

Ralf