Configure.pl allows sizeof(INTVAL) != sizeof(void *), but parrot requires them to be equal.

[doughera]

See  Original RT for more details.

On Linux/x86, the problem can be reproduced by running

perl Configure.pl --intval="long long int" --opcode="long long int"

[jkeenan]

Replying to doughera:

See  Original RT for more details. On Linux/x86, the problem can be reproduced by running

{{{ perl Configure.pl --intval="long long int" --opcode="long long int" }}}

I reproduced this tonight. Configure.pl told me:

auto::sizes -         Determine some sizes...
Hmm, I see your chosen INTVAL isn't the same size as your pointers.  
Parrot should still compile and run, but you may see a ton of warnings.
............................done.

... and I did indeed get a ton of warnings. But then Parrot failed to build:

./parrot -o runtime/parrot/library/PCT/HLLCompiler.pbc 
  compilers/pct/src/PCT/HLLCompiler.pir
/usr/local/bin/perl -e "" > compilers/pge/PGE/builtins_gen.pir
./parrot -o runtime/parrot/library/PGE.pbc compilers/pge/PGE.pir
./parrot runtime/parrot/library/PGE/Perl6Grammar.pir 
  --output=compilers/pge/PGE/builtins_gen.pir compilers/pge/PGE/builtins.pg
make: *** [runtime/parrot/library/PGE.pbc] Segmentation fault
make: *** Deleting file `runtime/parrot/library/PGE.pbc'
Finished

(Of course, the build failure may have been due to the inappropriateness of those command-line options for my box.)

So, would the proper course of action be to require that sizeof(INTVAL) == sizeof(void *)?

Thank you very much.

kid51

[jkeenan]

Replying to jkeenan:

Replying to doughera: [snip] So, would the proper course of action be to require that sizeof(INTVAL) == sizeof(void *)?

Can anyone shed any light on the issues raised in this ticket?

Thank you very much.

kid51

[doughera]

Can anyone shed any light on the issues raised in this ticket?

I suspect it could go either way. Either Configure.pl could be altered to require those sizes to be the same, or parrot could be altered to allow them to be different. I suspect sentiment leans towards altering Configure.pl, but I think it would be wise to check whether sizeof(long) == sizeof(void *) on all target platforms (including Win64) before proceeding.

[darbelo]

Replying to doughera:

Can anyone shed any light on the issues raised in this ticket?

I suspect it could go either way. Either Configure.pl could be altered to require those sizes to be the same, or parrot could be altered to allow them to be different. I suspect sentiment leans towards altering Configure.pl, but I think it would be wise to check whether sizeof(long) == sizeof(void *) on all target platforms (including Win64) before proceeding.

I agree that it can go either way and I think we should put some thought into making sure we pick the right way now, so we don't have to go back later. As an extra data-point, this is what auto::sizes has to say about this.

Hmm, I see your chosen INTVAL isn't the same size as your pointers.  Parrot
should still compile and run, but you may see a ton of warnings.

It's quite easy to miss in the midst of the Configure output, but to me, indicates the original intent clearly. Is this something we still want to support?

[jkeenan]

Replying to doughera:

I suspect it could go either way. Either Configure.pl could be altered to require those sizes to be the same, or parrot could be altered to allow them to be different. I suspect sentiment leans towards altering Configure.pl, but I think it would be wise to check whether sizeof(long) == sizeof(void *) on all target platforms (including Win64) before proceeding.

Any win64 users who could comment here?

Thanks.

kid51