Remove the special declaration for ml*. Instead, pass in the variable
to the functions that need it. It appears to be first defined in tbdu
and is used in getminor and nbn, so add ml* as a parameter for these
two functions.
It seems that `e` does not need to be declare to be a special variable
in sin.lisp. Removing the declaration does not cause any compiler
warnings/notes and the warning in limit.lisp about `e` being an
ignored special variable is gone now in clisp.
My guess is that there was some confusion because `e` is used to match
things via the m2 matcher. But it's not an actual variable; it's used
as a simple symbol to denote the item that was matched.
Possibly `w` and `y` are also used in this way. I didn't check.
sprdet.lisp declares `x` as a special variable and then undeclares it
later. However, `x` only appears to be used as a speical variable in
the function `nbn`. So add `x` as a parameter to `nbn` and remove the
special declarations for `x`.
As a test, we use `x` again in `simplim%limit` again to see there are
compiler warnings/notes. Neither clisp nor cmucl complain about
ignoring the special variable `x` any more.
Both cmucl and clisp warn that the variable `A` is a special variable
that is being ignored. To fix this, just rename the variable.
I am unable to find where `A` becomes a speical variable. This only
seems to happen when compiling maxima from scratch. If you compile
hypergeomtric.lisp by itself, there's no such warning.
02dfab3...
by
Wolfgang Dautermann <email address hidden>
Fix several tests with with_default_2d_display().
These tests (especially containing printf(..."~m"...))
did work in command line Maxima, but if a frontend like
wxMaxima was used, they failed - the ~m produced some XML (for
other frontends probably other code, as Gunter mentioned), which
caused the tests to fail.
Other tests already also use this 'protection'.
Now the testsuite and share_testsuite run without errors in
Maxima and wxMaxima (using Clisp).
