On Wed, 02 May 2001 13:30:06 EDT, Dan Sugalski wrote:
>At 09:58 AM 5/2/2001 -0700, Gurusamy Sarathy wrote:
>>On Wed, 02 May 2001 11:37:27 EDT, Dan Sugalski wrote:
>> >+++ perl.c Mon Apr 30 20:52:53 2001
>>[...]
>> >+# if defined(VMS)
>> >+ init_os_extras(aTHXo);
>> >+# else
>> > init_os_extras();
>> >+# endif
>>[...]
>> >+++ vms/vms.c Tue May 1 18:00:40 2001
>>[...]
>> > void
>> >-init_os_extras()
>> >+init_os_extras(pTHX)
>> > {
>> > char* file = __FILE__;
>> >- dTHX;
>>
>>I think it is better the way it was in those places. Avoiding dTHX
>>only makes any kind of sense for things that are hyper-performance
>>critical. IMO, keeping a uniform signature for init_os_extras()
>>is a loftier goal here.
>
>I could've sworn that it wasn't valid to do that before the first
>interpreter was fully set up, but I presume not?
But I don't see what that has anything to do with it--the patch hunks
above do not change any of the semantics. Instead of fetching the
interpreter from thread-local storage (which perl_alloc() initialized),
the change passes it in as an argument. The effect is the same.
Or is the real problem that INIT_TLS_AND_INTERP needs fixing on VMS?
Sarathy
[EMAIL PROTECTED]
