[H390-VM] IND$FILE and VM/SP

[[e-mail] [H390-VM]]

I tried mecafftools-on-diag58static-1.1.0.aws on VM/SP 5 (testing for the LCM+L machine), but it reports:


IND$FILE GET PROFILE EXEC A (ASCII CRLF

DMSITP141T Operation exception occurred at 035722 in routine IND$FI5S

I looked at rebuilding it (gcc from the sixpack 1.2 does run on a stock VM/SP system, hooray!), but there are things about the environment that make think I've got the wrong version or am looking at the wrong files or ...

Would someone with expertise in MECAFF's IND$FILE care to lend a hand? (Or just take the lead to build/debug the silly thing on VM/SP for the good of us all?)

If so, send me mail privately, swhobbit at derbyshire dot us.

Thanks!
Drew Derbyshire

[Ivan Warren [e-mail] [H390-VM]]

 IND$FILE GET PROFILE EXEC A (ASCII CRLF

*DMSITP141T Operation exception occurred at 035722 in routine IND$FI5S*

What instruction opcode failed ?

--Ivan

[[e-mail] [H390-VM]]

Hats off to Bob Polmanter (WABLY) who went through hoops to get a working IND$FILE on LCML4361 before I had even asked. He promptly contacted me, and the program is now publicly installed on the Y-DIsk.

-ahd-

[[e-mail] [H390-VM]]

Post by ***@kew.com [H390-VM]
I looked at rebuilding it (gcc from the sixpack
1.2 does run on a stock VM/SP system, hooray!),

GCC on sixpack 1.2 is very old. I suggest
using either the latest released version, or
I can give you a beta, as I have made a
number of bug fixes since last release.

Post by ***@kew.com [H390-VM]
but there are things about the environment
that make think I've got the wrong version or
am looking at the wrong files or ...

If you can show me an error in the running
of GCCCMS I may be able to diagnose the
problem.

BFN. Paul.

[[e-mail] [H390-VM]]

Post by ***@yahoo.com.au [H390-VM]
I can give you a beta, as I have made a
number of bug fixes since last release.

The latest beta is available here:

https://groups.yahoo.com/neo/groups/H390-VM/files/gcccms-stage286-exe.vmarc

Note that I am not sure if you need to build
against my PDPCLIB (above) or to Robert's
GCCLIB (which was never merged in to
PDPCLIB).

BFN. Paul.

[Drew Derbyshire [e-mail] [H390-VM]]

Post by ***@yahoo.com.au [H390-VM]

Post by ***@yahoo.com.au [H390-VM]
I can give you a beta, as I have made a
number of bug fixes since last release.

https://groups.yahoo.com/neo/groups/H390-VM/files/gcccms-stage286-exe.vmarc
Note that I am not sure if you need to build
against my PDPCLIB (above) or to Robert's
GCCLIB (which was never merged in to
PDPCLIB).

Bob Polmanter cleared the immediate problem (IND$FILE is now publicly
installed and working), but I suspect we'll need GCC on the system to
prevent user riots.  You suggest the beta for general public use?

Do you and Robert need to merge your libraries?

[[e-mail] [H390-VM]]

Post by Drew Derbyshire ***@kew.com [H390-VM]
Bob Polmanter cleared the immediate problem
(IND$FILE is now publicly installed and working),
but I suspect we'll need GCC on the system to
prevent user riots. You suggest the beta for general
public use?

Yes, I believe I have stopped development
now, so the beta that I uploaded is going to
be approximately what will be shipped in
the next release.

Post by Drew Derbyshire ***@kew.com [H390-VM]
Do you and Robert need to merge your libraries?

Robert had a comment in his code, where he
processed 1 byte at a time, and had a comment
that said "I promise to fix this". I explained to
him that if he fixed it he'd end up with the same
format as the existing record-oriented processing
in PDPCLIB, and so I believe he needed to fit
in with that. I think he agreed with me, but I
believe he's out of the game now, pursuing
other interests.

I don't know what the way forward is, but for
anyone who is happy to use normal PDPCLIB,
the way forward is to accept my GCCCMS
bug fixes while waiting for the formal release,
or at least use the latest formal release.

BFN. Paul.

['Dave Wade' [e-mail] [H390-VM]]

-----Original Message-----
Sent: 16 June 2018 00:47
Subject: Re: [H390-VM] Re: IND$FILE and VM/SP

Post by Drew Derbyshire ***@kew.com [H390-VM]
Bob Polmanter cleared the immediate problem (IND$FILE is now publicly
installed and working), but I suspect we'll need GCC on the system to
prevent user riots. You suggest the beta for general public use?

Yes, I believe I have stopped development now, so the beta that I uploaded
is going to be approximately what will be shipped in the next release.

Post by Drew Derbyshire ***@kew.com [H390-VM]
Do you and Robert need to merge your libraries?

I don't think its possible to merge the libraries. I believe that Roberts libraries use the native CMS interfaces to the file system and storage management.
This avoids issues with running other programs from "C" on VM where the OS storage management chains get reset when an OS program ends.
This is fine but it means you can't run MVS programs that access files by DD name and use FILEDEF.

Robert had a comment in his code, where he processed 1 byte at a time, and
had a comment that said "I promise to fix this". I explained to him that if he
fixed it he'd end up with the same format as the existing record-oriented
processing in PDPCLIB, and so I believe he needed to fit in with that. I think
he agreed with me, but I believe he's out of the game now, pursuing other
interests.
I don't know what the way forward is, but for anyone who is happy to use
normal PDPCLIB, the way forward is to accept my GCCCMS bug fixes while
waiting for the formal release, or at least use the latest formal release..

Do we have the source to Roberts Libraries? I couldn't find them anywhere to rebuild with the updated "C"?

BFN. Paul.

Dave

[[e-mail] [H390-VM]]

Post by 'Dave Wade' ***@gmail.com [H390-VM]
This is fine but it means you can't run MVS programs
that access files by DD name and use FILEDEF.

Is the problem that one MVS-style program can't call
another? Do we actually need that? The use of
DDNAMEs in PDPCLIB is arbitrary. It will happily
accept a filename instead.

Post by 'Dave Wade' ***@gmail.com [H390-VM]
Do we have the source to Roberts Libraries? I
couldn't find them anywhere to rebuild with the
updated "C"?

I've just uploaded the latest GCCLIB I have on
my hard disk. I don't know if that is the latest.
Did you look at the GCCCMS disks on the
unmodified sixpack 1.2?

I found the reference I mentioned:

// I promise to fix this code to not call fputc

Also in the files area I noticed that Hans
Latz has uploaded his own version of
GCCLIB which doesn't use the dostrans,
whatever that is, and assuming that is a
good thing.

BTW, what is the file type of the file on
the source code tape? It's fixed blocks
of 805 bytes and it looks like this:

C:\scratch\aaa3>hexdump out.dat | head
000000 02C3D4E2 40004761 5C5C5C5C 5C5C5C5C ***@.Ga\\\\\\\\
000010 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C5C5C5C \\\\\\\\\\\\\\\\
000020 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C5C5C5C \\\\\\\\\\\\\\\\
000030 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C5C5C5C \\\\\\\\\\\\\\\\
000040 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C610047 \\\\\\\\\\\\\a.G
000050 615C4040 40404040 40404040 40404040 a\@@@@@@@@@@@@@@
000060 40404040 40404040 40404040 40404040 @@@@@@@@@@@@@@@@
000070 40404040 40404040 40404040 40404040 @@@@@@@@@@@@@@@@
000080 40404040 40404040 40404040 40404040 @@@@@@@@@@@@@@@@
000090 40404040 405C6100 47615C40 40E38889 @@@@@\a.Ga\@@...
...

Is there a PC utility to convert that into
something more readable so that I don't
need to run VM?

BFN. Paul.

['Dave Wade' [e-mail] [H390-VM]]

-----Original Message-----
Sent: 16 June 2018 11:08
Subject: RE: [H390-VM] Re: IND$FILE and VM/SP

Post by 'Dave Wade' ***@gmail.com [H390-VM]
This is fine but it means you can't run MVS programs that access files
by DD name and use FILEDEF.

Is the problem that one MVS-style program can't call another? Do we actually
need that? The use of DDNAMEs in PDPCLIB is arbitrary. It will happily accept
a filename instead.

Post by 'Dave Wade' ***@gmail.com [H390-VM]
Do we have the source to Roberts Libraries? I couldn't find them
anywhere to rebuild with the updated "C"?

I've just uploaded the latest GCCLIB I have on my hard disk. I don't know if
that is the latest.

Thanks

Did you look at the GCCCMS disks on the
unmodified sixpack 1.2?

I could only find the object.

// I promise to fix this code to not call fputc
Also in the files area I noticed that Hans Latz has uploaded his own version of
GCCLIB which doesn't use the dostrans, whatever that is, and assuming that
is a good thing.

That’s a flag that CMS uses to decide if SVCs should be interpreted as DOS or OS calls (That’s not quite true but it will suffice) so using that flag broke some of the compilers that wanted DOS mode.

BTW, what is the file type of the file on the source code tape? It's fixed
C:\scratch\aaa3>hexdump out.dat | head
000010 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C5C5C5C \\\\\\\\\\\\\\\\
000020 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C5C5C5C \\\\\\\\\\\\\\\\
000030 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C5C5C5C \\\\\\\\\\\\\\\\
000040 5C5C5C5C 5C5C5C5C 5C5C5C5C 5C610047 \\\\\\\\\\\\\a.G
000050 615C4040 40404040 40404040 40404040
000060 40404040 40404040 40404040 40404040
@@@@@@@@@@@@@@@@
000070 40404040 40404040 40404040 40404040
@@@@@@@@@@@@@@@@
000080 40404040 40404040 40404040 40404040
@@@@@@@@@@@@@@@@
....

Basically the CMS file system uses 800 byte blocks on disk. This is just a dump of these blocks to tape.
There are two formats "TAPE DUMP" or "VMFPLC2".

Is there a PC utility to convert that into something more readable so that I
don't need to run VM?

There should be some utils about to do this.

https://groups.yahoo.com/neo/groups/H390-VM/files/PC%20Utilities/

has one version.

BFN. Paul.

Dave

------------------------------------
------------------------------------
------------------------------------
Yahoo Groups Links

[[e-mail] [H390-VM]]

Post by 'Dave Wade' ***@gmail.com [H390-VM]
There should be some utils about to do this.

Thanks. I was able to extract what I wanted with:

cmstape -x -f GCCLIBv06src.aws

The changes compared to PDPCLIB seem pretty
massive.

This probably needs to stay as an independent
product for the forseeable future.

Not sure if you want to switch to Hans's version too.

BFN. Paul.