as unsigned char just before passing isalnum() in ISWORD() rather than
fetching it as unsigned char. It is ok since EOF is never passed to ISWORD()
(though it will be a bit inefficient.)
Addresses PR#24929.
btw: the maximum length of an RE is limited by the "int" range, also
on the alpha. "sopno" is "int", and "int"s are used as counters within
the code, so it would need some work to change this. I only don't know
how one could test it...
sizeof(u_int32_t) != sizeof(*)), at least in cases with re->re_g->nstates
between 32 and 64.
Primary reason for the breakage was that the "states1" definition didn't
work as expected. (It didn't work before either, but this was not noticed
due to sizeof(long)==sizeof(*).)
The alpha can handle larger problems with the "small" state machine model
if a "long" is used as state variable, so it is better to keep the old
definition here. (u_int32_t is left for the "operator" variables.)
Use "int" as "sopno" - this is used as index into the states field, there
is no point in using a fized-size type.
!= 32 bits. This might need a bit more thought (i.e. why use long here to
begin with, when int32_t might be a better choice), but for now it's good
enough. From Ross Harvey in PR 3450, extended by me to cover the rest of
the opcode constants and masks.
