Leo wrote:
>>Its seems to be the #ifdef vs #if issue.
In theory that issue should only occur to the JIT_CAPABLE
defines and not to the HAVE_COMPUTED_GOTO ones,
I seem now. Because we have three places in the Configure
script where we define
-DHAS_JIT in jit.pl,
#define PARROT_JIT_CAPABLE ${jitcapable} in configure_h.in
#define JIT_CAPABLE ${jitcapable} in configure_h.in again
(I can't think of one reason or other why we have those three)
versus one place which I found in cgoto.pl where we set up
-DHAVE_COMPUTED_GOTO
So we shoudn't have any appearance of that issue, as we doesn't
ever do following
#define HAVE_COMPUTED_GOTO ${cgoto}
Or we do? But where then? I can't find such a place, nohow.
And what is the reason that we have
#if HAVE_COMPUTED_GOTO
in exec.c, interpreter.c, jit2h.pl now?
Leo wrote:
>Compiles now but:
>$ make test
>..
># got: 'Computed goto unavailable in this configuration.
># '
I tried to build and to test Parrot fixed with the patch I
provided on Windows 2000(building with MSVC), and on Madrake
Linux 9.0 (building with gcc) and it was all fine. I passed
--cgoto=0 and --cgoto=1 into the configure script and coudn't
cause "Computed goto unavailable in this configuration". So
I can't track down and solve that problem. That's why I propose
to split the patch into three pieces
a piece which wants to deal with GC_IS_MALLOC
a piece which wants to deal with HAVE_COMPUTED_GOTO
and a piece which wants to deal with HAS_JIT
and try to apply those separately.
I plan to prepare patches for the 1st and the 2nd piece and leave
the 3rd piece untill we will see a design desicion on "wether to use
defined/undefined or 0/1 define sheme"
