develooper Front page | perl.perl5.porters | Postings from February 2008

Re: Partial fix for change 33291 to get VMS building again.

Thread Previous
Craig A. Berry
February 22, 2008 08:20
Re: Partial fix for change 33291 to get VMS building again.
Message ID:
At 7:55 PM -0600 2/19/08, John E. Malmberg wrote:
>Craig A. Berry wrote:
>>At 5:42 PM -0600 2/18/08, John E. Malmberg wrote:
>>>proto.h needs to expose some macros on VMS that it does not after change 33291.
>>>I could not figure out how to generate a proto.h that looks like
>>>the  one produced by the attached patch.
>>Let me take a stab at getting embed.fnc or whatever to work. We'll
>>hold off on applying your patch for now, but yes, something needs to be
>>done to get things buildable again.

Here's what I ended up doing:

>>>When I changed proto.h to expose those macros, it also exposed some
>>>function prototypes that were different than in the VMS specific code.

>In vms.c, it is removing the casts which are not needed when using the same prototypes as the win32 variant.

I kept that part and changed the prototype of the internal
do_spawn2() to match.

>The only reason that I can see to have the different prototypes is if the compiler for that platform does not support them.  In this case all versions of DECC should support those prototypes.

And the main reason to have them the same is that it also gives us
the new assert macros.  To maintain separate prototypes we'd now have
to maintain separate macros as well.

>It appears that win32 has a variant of do_spawn that does not wait for completion.  I am short of cycles to see what the impact of that being missing on VMS is right now.

I added Perl_do_spawn_nowait(), which is almost copy-and-paste from
the Win32 routine of the same name.

Craig A. Berry

"... getting out of a sonnet is much more
 difficult than getting in."
                 Brad Leithauser

Thread Previous Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About