On Wed, 2 Nov 2016 14:02:42 -0500, Kirk Wolf wrote:
>
>Had you seen this?
>It appears that VM/CMS OpenExtensions folks have thought a little more
>about the conversion of fork/exec to spawn:
>
>https://www.ibm.com/support/knowledgecenter/SSB27U_6.2.0/com.ibm.zvm.v620.dmsa3/forkcnv.htm
>
Well, that's because OpenExtensions doesn't have a working fork().
"Necessity is the mother ..." It has something they (dishonestly)
call fork, but it's more like what (older?) versions of UNIX call "vfork()"
And there's posix_spawn, for which the rationale is to provide support for
hardware lacking an MMU. It's incompatible with and somewhat inferior
to IBM's spawn because a design criterion was that it should be possible
to implement posix_spawn as a wrapper to fork on systems on which
true fork is available.
And what it's truly missing is a way to do a fork() *without* the exec().
I've wished for that in Rexx on z/OS.
>On Wed, Nov 2, 2016 at 11:14 AM, John McKown wrote:
>>
>> Correct. I was actually looking at this when I was doing my port of BASH
>> for the CBT. But I've basically abandoned my port since Rocket Software
>> took over the ported tools from IBM. I figured that they'd do a better job
>> than I. BASH does some really strange things to run commands. It's not a
>> simple fork()/exec() sequence which would be relatively simple to replace
>> with a spawn(), which is required to honor _BPXAS_SHAREAS.
>>
Of course sh provides some of that facility by reading scripts with the
dot (".") command which runs in the same execution environment.
I'd like Rexx added to the mix so I could somehow invoke BPXWDYN
from a shell script. Perhaps a shell built-in interface to ATTACH?
Interestingly, if I invoke a Rexx EXEC in TSO that does SYSCALL chdir,
that CWD is set for the remainder of the TSO session. No surprise.
A while back, I tried to do some Apache work (port to z/OS; IBM
beat me by a nose). Apache rejected one of my build script patches
that would have greatly improved build performance on z/OS because
it depended on a POSIX construct not supported by Bourne Shell.
Yet they required an ANSI-conformant C compiler and now depend
on bash?
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
