[Axiom-developer] interpsys and algebra

[root]

Camm,

I think you're running into limitations due to the interactions
of the hand-built image and your interpsys image. You really need
the fully compiled algebra with the new patches which does not yet
exist. I'm still working on that level of build machinery.

You may be able to use your new interpsys, install it into the
hand-built directory, and recompile all of the algebra files 
that are used in this example. You can figure out which algebra
files are used by running an example, looking for messages like:
   Loading /fix/t1/camm/axiom/noncvs/axiom/mnt/linux/algebra/URAGG.o 
and then at an axiom prompt typing:
)show URAGG
and it will tell you the source spad file to compile.

Tim
address@hidden
address@hidden

======================================================================
Greetings!  Here are my resulst thus far:

1) Copying the linux binary from tenkan, and using my compiled
   interpsys in that directory, I get a slightly modified result from
   the one reported:

        (3) -> set [p,p]
   Loading /fix/t1/camm/axiom/noncvs/axiom/mnt/linux/algebra/URAGG-.o 
      for domain UnaryRecursiveAggregate& 
   There are no exposed library operations named elt but there are 51 
      unexposed operations with that name. Use HyperDoc Browse or issue
                               )display op elt
      to learn more about the available operations.
 
   Cannot find a definition or applicable library operation named elt 
      with argument type(s) 
        List MonoidRing(Polynomial PrimeField 5,Permutation Integer)
      
      Perhaps you should use "@" to indicate the required return type, 
      or "$" to specify which version of the function you need.
(3) -> one? p

   Loading /fix/t1/camm/axiom/noncvs/axiom/mnt/linux/algebra/BOOLEAN.o 
      for domain Boolean 
   (3)  false
                                                                Type: Boolean

2) Taking the same interpsys, copying over the missing algebra clisp
   files from Juergen's tree, compiling them by hand, and then
   executing the above, all works correctly.

3) These two results with safety set to 0.   My current understanding
   is that the large argument patch is only necessary when compiling
   with safety >=2, as GCL will inline the large apply call otherwise.
   In any case, the |data| patch for define.boot.pamphlet has been
   applied. 

4) Logical next step is to diff Juergen's clisp files with the ones
   pertaining to the binary.  These alas are not in the tarball.

5) Are we sure we know this problem persists in a clean system
   recompiled with the lengthvec patch?  I currently cannot reproduce
   it from source, only partially with some binary only modules.

6) Does everyone else see this elt message in 1)?  Perhaps this
   indicates a syntax error or is otherwise illustrative?