Probably not countable.

--- bittest(x:small,y:small) is wrong if y>31

--- x:small%y:small cannot use % in C because -2%3 would be wrong

Things that are poorly implemented:
----------------------------------

--- Array (vector, matrix) may be disconnected in the stack.

--- vector() and matrix() do not garbage collect.

--- Garbage collecting should always be done after a loop.

--- Code generated for "a=if(b,c,d)" constructs is ugly.

--- user-defined member functions should be inlined

Things that are not implemented:
--------------------------------

--- a[b,]+=c

--- issquare(4,&a[b,])

--- Numerical derivative because it is static to anal.c

--- Embedded function definitions. f(x)=g(y)=x+y

--- Complex statement like sumalt, etc... No interface in the library.

--- Argument reduction like 1+I*3-->complex(1,3). I have the code but I look
    for a nice representation in C.

Discrepancy between gp2c and gp:
--------------------------------

--- =\n= is allowed by gp not by gp2c. Don't cry.

--- Comments after split-line stray \ are not accepted. Ouch.

--- Spaces in middle of term like (a b==ab) are not allowed. Be sensible.

--- [GP 2.1] code generated for matrix(n,p,i,i/*sic*/,f(i)) will not compile.

--- gp2c does not allow functions named local() to be called.

--- [GP 2.1 only] gp2c does not allow : as separator.

--- "error" output PARI errors, not user errors.

--- gp2c is generally more lenient than gp about syntax.

--- Do not catch misuse of GP 'global()' vars.

--- Scope of variable after function call are not the same in GP and
    in C. Seem very problematic.

--- Use of small may lead to overflow problems.

--- [GP 2.1 only] Member functions works only with unambiguous typed args. 

--- return() can return 0 instead of nothing.
