On 2012-06-20 17:13, McKown, John wrote:
-----Original Message-----
From: IBM Mainframe Discussion List
[mailto:[email protected]] On Behalf Of Gord Tomlin
Sent: Wednesday, June 20, 2012 3:30 PM
To: [email protected]
Subject: Re: Calling idcams
One could argue that creating LE enabled programs would not be very
"standard". You'd be challenged to find any UNIX implementation that
could run LE enabled programs, other than the one we try to avoid
calling USS!
OK, I get your meaning. But for all main HLLs on z/OS (COBOL, FORTRAN, PL/I, IBM C/C++), it is
"standard" to use LE. Just like on Linux, I'd say that it is "standard" to use
a glibc compatible environment.
>>
>> From a GCC perspective, one would normally expect to produce
>> programs
>> that operate with GNU's library.
>
> True. What might be nice, on z/OS, is to implement a library of
"stubs" which are glibc compatible, but map the glibc calling sequence
into the associated LE subroutine's calling sequence. Well, at least
__I__ think it would be nice.
Using GCC on z/OS is a collision of technologies that were not designed
to work together. GCC is designed with glibc in mind, and IBM XL C/C++
is designed with the LE runtime in mind.
If GCC and the LE runtime are to inter-operate, then something has to
give. If I was someone porting an app developed in GCC on another
platform, then I might be hoping that it would use glibc and run just
like it did on the other platform. On the other hand, if I (like you)
was someone wanting to do do new C development on z/OS, then I might
prefer it if the compiler output could talk to the LE runtime. Needless
to say, IBM would prefer to see the first option above rather than the
second!
I actually rather like the CEEDUMP output. It is well known that I am
weird.
To each his own on that one...I try to ensure that I always get a "real"
dump instead of a CEEDUMP, and I eliminate LE's ESPIE whenever possible
as well.
--
Regards, Gord Tomlin
Action Software International
(a division of Mazda Computer Corporation)
Tel: (905) 470-7113, Fax: (905) 470-6507
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
