CoCoA » CoCoALib

See RingHom.[HC]

  RingElem PartialRingHom::operator()(ConstRefRingElem x) const

Added phi(matrix).
Prototype. Not entirely certain with it.
CVS'd

I have a prototype for list/vector; not checked, not checked-in.
Should I check it in? Would you test it?

I have updated ValgrindTests.sh --now it should give a helpful error mesg when valgrind is not installed (or not in your PATH).
Checked in!

Implemented for vector also in CoCoA-5, where it'll be most useful.
It was much more painful than expected because of the old style on RINGHOM::eval in Interpreter.C.
At the end I've rewrittein it using evalArgAsT1orT2orT3orT4orT5orT6, switch, Value::from, RefTo... and it seems to work fine.
I also added a test in bug-EvalTwice.cocoa5 (OK)

The manual entry for "Intro to RingHom" needs to be updated.

Must also add some good tests, e.g.

phi := IdentityHom(R);
phi([1,2,3]);  -->  ***!!!CRASH!!!***  But should work, right?

I am not sure whether the following behaviour is what is wanted:

phi := CanonicalHom(R);
phi([1,2,x]); --> works as one might hope; x is indet in R
phi([1,2,3]); --> ERROR: Argument to RingHom must be in domain

Should the second really give error?

We should move towards making apply obsolescent in CoCoA-5 now that RINGHOM can be applied directly to LIST and MAT.

Many packages use apply, and several tests use apply too.

We should update the doc so that it encourages people to apply RINGHOM directly rather than using apply.
We should also gradually removed uses of apply from the packages.

And, as written in the previous comment, make apply obsolescent.

I have now removed apply from all the packages as well... sigh!
All tests pass now.

This issue overlaps with Design #1467.