On Wed, Jan 21, 2015 at 9:56 AM, Paul Gilmartin <
[email protected]> wrote:

> On Wed, 21 Jan 2015 09:40:22 -0600, John McKown wrote:
> >
> >The most difficult part has already been done because there is a LLVM
> >variant for Linux on z. The porting would be to make this variant compile
> >and run using IBM's XLC instead of GCC.
> >
> The difficulty metric as I perceive it, graphically:
>
>      GCC -------> GCC ----------------------------------------> XLC
>     ASCII        ASCII                                         EBCDIC
>     Intel          z                                             z
> little-endian  big-endian                                    big-endian
>
> (But then, I hate EBCDIC!)
>
> --gil
>

​Depending on the actual code base, the ASCII -> EBCDIC might be simple,
like it was for SQLite (due to Dr. Hipp's excellent technique). Or somewhat
difficult, as it was for the embedded "GNU readline" in BASH. But if LLVM
assumes a correspondence between the ASCII character and it's internal
representation (such as using \0x0A instead of \r for a "new line
character"), then it will be a real problem.

Of course, the real problem would be any dependency on other GNU software
such as GNU's version of M4, gawk, or sed which is not available on z/OS.
I've run into that sort of thing in the configuration scripts (not the
actual project code) before.

Of course, what I _could_ do (joking) is to port Hercules/390 to run under
z/OS UNIX and then run z/Linux under Hercules/390 on z/OS. What would be
really weird would be to look at the newly released KVM for z/Linux which
turns z/Linux into a hypervisor, as it can be on Intel, and make that
"somehow" work under z/OS. And I can imagine Peter Relson having a heart
attack about now, thinking about the holes this could put in z/OS
reliability and security.

Actually, maybe it would be better to write a program, like WINE, which can
load z/Linux ELF executable programs, with "stubs" for the z/Linux glibc
API to redirect to the XLC runtime. This would allow me to do any required
code point translations, leaving the z/Linux to run in ASCII internally.
This would be difficult, but it might actually be a reasonable approach.
For someone with sufficient time and talent. Well, time for a "Boom Beach"
break! Which shows that I don't have the time. Gotta get "Dr. Terror".


-- 
​
While a transcendent vocabulary is laudable, one must be eternally careful
so that the calculated objective of communication does not become ensconced
in obscurity.  In other words, eschew obfuscation.

111,111,111 x 111,111,111 = 12,345,678,987,654,321

Maranatha! <><
John McKown

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to