First post: THANK YOU! and a question or two

[blackiron79]

Hello to the group,

I wanted to say a huge THANK YOU to Roger Bowler, Jay Maynard, Jay Moseley,
Volker Bandke, Tommy Sprinkle, and all the people in the Hercules and MVS
3.8J community.

I started coding on this very OS (and MFT running on another machine) and
it was such a huge thrill to see the system come up and run a few jobs
today. I've been using ESA facilities and 31 bit addressing for so many
years now it will be a real struggle to code in MVS/SP once again. But it
should be fun.

I am an assembler coder, and I've never been an MVS sysprog or operator, so
many parts of this have a steep learning curve for me.

I installed TK3 years ago but didn't do much with it. This time I went
through Jay Moseley's excellent instructions and built the system according
to his program. My system comes up and runs fine, and I tested a few small
jobs.

I want to add some of the compilers on Jay Moseley's site. Most of the JCL
requires SYS2.PROCLIB, so I created an IEFBR14 job and allocated
SYS2.PROCLIB using the DCB from SYS1.PROCLIB. However, as you are probably
thinking right now, SYS2.PROCLIB isn't in the proclib search so the test
job didn't find the PL/I proc.

Did I do anything wrong by using SYS1.AMASTCAT? Is there anything special
to do about creating SYS2.PROCLIB?

How do I add proclibs to the search order? I tried a /*PROCLIB card but it
wasn't recognized. I imagine it won't work for another 10 years or so!

For those of you who are running your system according to Jay's
instructions, what other modifications are necessary to install and use the
compilers properly?

Thanks to everyone!



               
___________________________________________________________
Now you can scan emails quickly with a reading pane. Get the new Yahoo! Mail. http://uk.docs.yahoo.com/nowyoucan.html

[Gerhard Postpischil]

Black Iron '79 wrote:
> I want to add some of the compilers on Jay Moseley's site. Most of the JCL
> requires SYS2.PROCLIB, so I created an IEFBR14 job and allocated
> SYS2.PROCLIB using the DCB from SYS1.PROCLIB. However, as you are probably
> thinking right now, SYS2.PROCLIB isn't in the proclib search so the test
> job didn't find the PL/I proc.

If you look in SYS1.PROCLIB, edit member JES2. You'll find a DD
card labeled PROC00. This card,and the DDs concatenated behind
it, defines the libraries the will be used to resolve PROC
lookups. If your new library has a blocksize that's larger than
any you currently have, add a DCB=BLKSIZE=32720 (or whatever) on
the PROC00 DD.

> Did I do anything wrong by using SYS1.AMASTCAT? Is there anything special
> to do about creating SYS2.PROCLIB?

For SYS1 and SYS2 (and some others) the master catalog is it.
For new high-level indices it is recommended that you put a user
catalog, and your data sets, on a new disk pack that won't be
affected when/if you upgrade your system.



Gerhard Postpischil
Bradford, VT

[blackiron79]

--- In H390-MVS@..., Gerhard Postpischil <gerhard@...> wrote:

> If you look in SYS1.PROCLIB, edit member JES2. You'll find a DD
> card labeled PROC00. This card,and the DDs concatenated behind
> it, defines the libraries the will be used to resolve PROC
> lookups. If your new library has a blocksize that's larger than
> any you currently have, add a DCB=BLKSIZE=32720 (or whatever) on
> the PROC00 DD.

Thank you, I'll try it. The DCBs are the same for both SYS1 and SYS2.PROCLIB.

> For SYS1 and SYS2 (and some others) the master catalog is it.
> For new high-level indices it is recommended that you put a user
> catalog, and your data sets, on a new disk pack that won't be
> affected when/if you upgrade your system.

Thanks for the info!

Jack

[blackiron79]

--- In H390-MVS@..., "blackiron79" <blackiron79@...> wrote:

>
> --- In H390-MVS@..., Gerhard Postpischil <gerhard@> wrote:
>
> > If you look in SYS1.PROCLIB, edit member JES2. You'll find a DD
> > card labeled PROC00. This card,and the DDs concatenated behind
> > it, defines the libraries the will be used to resolve PROC
> > lookups. If your new library has a blocksize that's larger than
> > any you currently have, add a DCB=BLKSIZE=32720 (or whatever) on
> > the PROC00 DD.
>

I just checked and I don't have a member named JES2 in my SYS1.PROCLIB. I'm not using the TK3 system, I'm using Jay Moseley's build instructions. Should I have a copy of this job somewhere on my system, or do I need to find a copy and add it?

Thank you.

[blackiron79]

Sorry for replying to myself, I couldn't edit my msg.

I may not be understaning how TSO edit works, I tried to use it to look at the contents of SYS1.PROCLIB and it went into create mode for the JES2 member.

in TSO:

edit sys1.proclib(jes2) cntl

DATA SET OR MEMBER NOT FOUND, ASSUMED TO BE NEW
INPUT
00010

then I hit enter to get into edit mode and then typed end to exit. I thought this meant that the JES2 member didn't exist.

But then on a hunch I tried to edit sys1.maclib(getmain) asm and it gives me the same results. I know GETMAIN exists in SYS1.MACLIB because I just submitted an assembly and SYS1.MACLIB is the only SYSLIB data set and the assembly finds the GETMAIN and expands it.

I have my profile noprefix and I'm using ibmuser with a logon size of 4000.

Any ideas why nothing seems to exist?

[Binyamin Dissen]

On Wed, 12 Aug 2009 12:57:26 -0000 "blackiron79" <blackiron79@...>
wrote:

:>Sorry for replying to myself, I couldn't edit my msg.

:>I may not be understaning how TSO edit works, I tried to use it to look at the contents of SYS1.PROCLIB and it went into create mode for the JES2 member.

:>in TSO:

:>edit sys1.proclib(jes2) cntl

:>DATA SET OR MEMBER NOT FOUND, ASSUMED TO BE NEW
:>INPUT
:>00010

:>then I hit enter to get into edit mode and then typed end to exit. I thought this meant that the JES2 member didn't exist.

:>But then on a hunch I tried to edit sys1.maclib(getmain) asm and it gives me the same results. I know GETMAIN exists in SYS1.MACLIB because I just submitted an assembly and SYS1.MACLIB is the only SYSLIB data set and the assembly finds the GETMAIN and expands it.

:>I have my profile noprefix and I'm using ibmuser with a logon size of 4000.

:>Any ideas why nothing seems to exist?

EDIT adds CNTL or ASM as a qualifier. Fully qualify the name.

--
Binyamin Dissen <bdissen@...>
http://www.dissensoftware.com

Director, Dissen Software, Bar & Grill - Israel


Should you use the mailblocks package and expect a response from me,
you should preauthorize the dissensoftware.com domain.

I very rarely bother responding to challenge/response systems,
especially those from irresponsible companies.

[Greg Price-3]

On Wed, 12 Aug 2009 16:06:33 +0300, Binyamin Dissen wrote:

>EDIT adds CNTL or ASM as a qualifier. Fully qualify the name.

Exactly!  While PROF NOPREF means that it takes the first
qualifier from the supplied dsname (in this case SYS1),
because the supplied dsname is not quoted EDIT adds the
low level qualifier of the type specified (in this case CNTL).

As was reported, no SYS1.PROCLIB.CNTL or SYS1.MACLIB.CNTL
already exists.

Cheers,
Greg

[kerravon86]

--- In H390-MVS@..., Black Iron '79 <blackiron79@...> wrote:
>
> I started coding on this very OS (and MFT running on
> another machine) and it was such a huge thrill to see
> the system come up and run a few jobs today.

Good to see you've got the bug. :-) Hopefully you'll
never shake it this time. :-)

> I've been using ESA facilities

Which particular ESA facilities are you thinking
of? "review" is available for ISPF-edit-like
facilities. etc etc

> and 31 bit addressing

That's available too from 2 places.

> I installed TK3 years ago but didn't do much with it.
> This time I went through Jay Moseley's excellent
> instructions and built the system according
> to his program. My system comes up and runs fine, and
> I tested a few small jobs.

JM's instructions make you want to do more than
you do with TK3?! I wish I had such inspiring
literary skills.

> How do I add proclibs to the search order? I tried a
> /*PROCLIB card but it wasn't recognized. I imagine
> it won't work for another 10 years or so!

Don't be too confident about things like that. In
another group someone said that txt2pdf wouldn't
ever be able to be run on the free VM/CMS and it
was working in about a week.

On the MVS side, it was said it wasn't possible to
intercept GETMAIN because it was the wrong sort of
SVC.

C:\mvs380\source\misc>wc svc120i.jcl
  966  5547 65435 svc120i.jcl

Less than 1000 lines of assembler says otherwise. :-)

BFN.  Paul.

[blackiron79]

--- In H390-MVS@..., Binyamin Dissen <bdissen@...> wrote:
> EDIT adds CNTL or ASM as a qualifier. Fully qualify the name.

Thanks. I tried a few things but it always required a type. If I specify edit 'sys1.proclib(jes2)' it asks for a type so I answered "text" and it worked. Not sure how to fully qualify the DSN other than quoting so that is all I tried.

In Jay Moseley's TSO Tutorial page the description doesn't explain what's going on here. It seems to suggest the type is for line numbering not as a DSN qualifier.

If I can find an explanation of this (or if someone else already knows) in an IBM manual, I'll post an update here.

[blackiron79]

--- In H390-MVS@..., "kerravon86" <kerravon86@...> wrote:
> Good to see you've got the bug. :-) Hopefully you'll
> never shake it this time. :-)

Thanks Paul!
>
> > I've been using ESA facilities
>
> Which particular ESA facilities are you thinking
> of? "review" is available for ISPF-edit-like
> facilities. etc etc

I mean I use the ESA and XA facilities as a coder. The linkage stack, PC routines, various ASC modes, the list goes on. Not sure when RESMGR was added, I think it was also ESA.

I'd really like to be able to write the same type of software at home as I do at work, but that's obviously not happening any time soon until somebody can get IBM to be nice to us.

>
> > and 31 bit addressing
>
> That's available too from 2 places.

I saw one place and while it looks good for certain apps it won't do yet for systems software development.

>
> > I installed TK3 years ago but didn't do much with it.
> > This time I went through Jay Moseley's excellent
> > instructions and built the system according
> > to his program. My system comes up and runs fine, and
> > I tested a few small jobs.
>
> JM's instructions make you want to do more than
> you do with TK3?! I wish I had such inspiring
> literary skills.

I just figured to try to do it the next harder way, but Jay made it too easy!

> On the MVS side, it was said it wasn't possible to
> intercept GETMAIN because it was the wrong sort of
> SVC.

I don't think GETMAIN is any harder than usual to intercept, not sure what the issue(s) were.

> BFN.  Paul.

Cheers,

Jack

[au1john]

> in TSO:
>
> edit sys1.proclib(jes2) cntl
>
> DATA SET OR MEMBER NOT FOUND, ASSUMED TO BE NEW
> INPUT
> 00010
>
> then I hit enter to get into edit mode and then typed end to exit. I
> thought this meant that the JES2 member didn't exist.
>
> But then on a hunch I tried to edit sys1.maclib(getmain) asm and it gives
> me the same results. I know GETMAIN exists in SYS1.MACLIB because I just
> submitted an assembly and SYS1.MACLIB is the only SYSLIB data set and the
> assembly finds the GETMAIN and expands it.
>
> I have my profile noprefix and I'm using ibmuser with a logon size of

4000.
>
> Any ideas why nothing seems to exist?

Using TSO EDIT requires understanding of the syntax of the EDIT command, how
it uses its operands and the way in which dataset names are handled in
native TSO. Single quotes (') are significant. E.g. E 'SYS1.PROCLIB(JES2)'
CN.  However you will then have ENQ problems.

Did you get to Jay's instructions for installing RPF? If not, I suggest you
do so. It will make things a little more familiar, similar to ISPF.

(I think TK3 MVS came with SYS2.PROCLIB added to JES2 and some compilers. It
also has an earlier version of RPF that you should upgrade using Jay's
instructions)

[Tony Harminc]

2009/8/12 blackiron79 <blackiron79@...>:
> --- In H390-MVS@..., Binyamin Dissen <bdissen@...> wrote:
>> EDIT adds CNTL or ASM as a qualifier. Fully qualify the name.
>
> Thanks. I tried a few things but it always required a type. If I specify edit 'sys1.proclib(jes2)' it asks for a type so I answered "text" and it worked. Not sure how to fully qualify the DSN other than quoting so that is all I tried.
>
> In Jay Moseley's TSO Tutorial page the description doesn't explain what's going on here. It seems to suggest the type is for line numbering not as a DSN qualifier.
>
> If I can find an explanation of this (or if someone else already knows) in an IBM manual, I'll post an update here.

The type keyword controls a number of EDIT behaviours. A type TEXT
will be edited in mixed case, whereas CNTL will be upper case. And the
type controls the attributes when you are creating a dataset, though
in general it will edit any shape or size dataset with any type
keyword without problem.

The overall scheme is a little like Windows' hiding of filetypes
("extensions") in Windows Explorer. The idea is that you would
typically create a new dataset with something like:
edit test(newmemb) asm
which would create 'userid.test.asm(newmemb)', and then you would
assemble it with
asm test(newmemb)
and the language prompter would know to use the appropriate trailing qualifier.

But the whole thing is not so easy to use when you are dealing with
naming schemes that IBM didn't plan for.

Tony H.

[blackiron79]

--- In H390-MVS@..., "au1john" <au1john@...> wrote:
> Using TSO EDIT requires understanding of the syntax of the EDIT
> command, how it uses its operands and the way in which dataset
> names are handled in native TSO.

Clearly. But we all had to learn at some point, didn't we? That's what I'm trying to do now, going by Jay's site. Punching cards actually seems easier now, now where did I put my 029?

> Did you get to Jay's instructions for installing RPF? If not, I
> suggest you do so. It will make things a little more familiar, ?
> similar to ISPF.

As long as I'm running like the good ole days, I thought it would be valuable to use the learn and use the native tools and facilities. I don't plan to install any of the extras on this system.

Thanks.

[blackiron79]

--- In H390-MVS@..., Tony Harminc <tharminc@...> wrote:
> The type keyword controls a number of EDIT behaviours. A type TEXT
> will be edited in mixed case, whereas CNTL will be upper case. And
> the type controls the attributes when you are creating a dataset,
> though in general it will edit any shape or size dataset with any
> type keyword without problem.

Thanks Tony. I made sure to set my caps lock on but I'm sure I'll learn more about this as I use it.

> The overall scheme is a little like Windows' hiding of filetypes
> ("extensions") in Windows Explorer. The idea is that you would
> typically create a new dataset with something like:
> edit test(newmemb) asm
> which would create 'userid.test.asm(newmemb)', and then you would
> assemble it with
> asm test(newmemb)
> and the language prompter would know to use the appropriate
> trailing qualifier.
>
> But the whole thing is not so easy to use when you are dealing with
> naming schemes that IBM didn't plan for.

It sounds odd that they would add that parameter as a required parameter of edit, it just seems to make things more complicated.

Thanks for your explanation, it helps.

Jack

[au1john]

>
> As long as I'm running like the good ole days, I thought it would be
> valuable to use the learn and use the native tools and facilities. I don't
> plan to install any of the extras on this system.
>

OK

A lot of the information has been moved to an appendix, but the current z/OS
1.10 TSO/E users guide can be used for MVS 3.8 TSO.

Here you will find it.
http://publibz.boulder.ibm.com/epubs/pdf/ikj4c240.pdf

I guess you would probably prefer what is at bitsavers  

http://www.bitsavers.org/pdf/ibm/370/OSVS2/GC28-0646-3_OSVS2_TSOcmd.pdf


I'd suggest, because of IBM's insistence on backward compatibility, you
wouldn't find anything in the document hosted on bitsavers that would not
work on z/OS 1.10.

[somitcw-2]

--- In H390-MVS@...,
 "blackiron79" <blackiron79@...> wrote:

> --- In H390-MVS@...,
>"kerravon86" <kerravon86@> wrote:
>> Good to see you've got the bug. :-) Hopefully you'll
>>never shake it this time. :-)
> Thanks Paul!
>>> I've been using ESA facilities
>> Which particular ESA facilities are you thinking
>>of? "review" is available for ISPF-edit-like
>>facilities. etc etc
> I mean I use the ESA and XA facilities as a coder.
>The linkage stack, PC routines, various ASC modes,
>the list goes on. Not sure when RESMGR was added,
>I think it was also ESA.
> I'd really like to be able to write the same type of
>software at home as I do at work, but that's obviously
>not happening any time soon until somebody can get IBM
>to be nice to us.

   It is slowly happening.   No ISPF/PDF yet but
FSE, RPF, REVIEW, QUEUE, and others are available.
System upgrades are available to add some TSO/E
CLIST support, VTAM terminal enhancements, some
disk support for 2311, 3375, 3380, 9345, and 3390.
Don't hold your breath waiting for IBM to do
something.

>>> and 31 bit addressing
>> That's available too from 2 places.
> I saw one place and while it looks good for certain
>apps it won't do yet for systems software development.

   It's a start.

>>> I installed TK3 years ago but didn't do much with it.

   Years of updates are available.

>>>This time I went through Jay Moseley's excellent
>>>instructions and built the system according
>>>to his program. My system comes up and runs fine, and
>>>I tested a few small jobs.
>> JM's instructions make you want to do more than
>>you do with TK3?! I wish I had such inspiring
>>literary skills.
> I just figured to try to do it the next harder way,
>but Jay made it too easy!
>> On the MVS side, it was said it wasn't possible to
>>intercept GETMAIN because it was the wrong sort of
>>SVC.
> I don't think GETMAIN is any harder than usual to
>intercept, not sure what the issue(s) were.

   Not all SVCs can be intercepted with all methods
( SVC Filtering, SVCTABLE update, ZAP, LPALIB load
module replacement, LPALIB module replacement link,
NUCLEUS relink, etc. ) but having a GETMAIN SVC
that is impossible to intercept does sound far out.

>> BFN.  Paul.
> Cheers,
> Jack

[Gerhard Postpischil]

somitcw wrote:
>    Not all SVCs can be intercepted with all methods
> ( SVC Filtering, SVCTABLE update, ZAP, LPALIB load
> module replacement, LPALIB module replacement link,
> NUCLEUS relink, etc. ) but having a GETMAIN SVC
> that is impossible to intercept does sound far out.

When I did the intercept, I was looking for a clean way to hook
task or region termination; unfortunately all the termination
code I found uses branch entry (in some cases undocumented).

Gerhard Postpischil
Bradford, VT

[somitcw-2]

--- In H390-MVS@...,
 Gerhard Postpischil <gerhard@...> wrote:

> somitcw wrote:
>> Not all SVCs can be intercepted with all methods
>>( SVC Filtering, SVCTABLE update, ZAP, LPALIB load
>>module replacement, LPALIB module replacement link,
>>NUCLEUS relink, etc. ) but having a GETMAIN SVC
>>that is impossible to intercept does sound far out.
> When I did the intercept, I was looking for a clean
>way to hook task or region termination; unfortunately
>all the termination code I found uses branch entry
>(in some cases undocumented).
> Gerhard Postpischil
> Bradford, VT

   Are you still looking for something?
IMS batch regions ending need to cleanup LSQA.
IMS uses DFSMRCL0 from hook:
NAME IGC0001C IEAVTRML
VER 00 0000000000000000
REP 00 C4C6E2D4D9C3D3F0  DFSMRCL0
It might work but if your cleanup routine is
ever missing from LPALIB/LST, IPL will fail.
Been there and IPL did fail because an IMS
sysprog deleted DFSMRCL0 from IMS.RESLIB so
MLPA couldn't find it.  IMS.RESLIB wasn't on
sysres so:
Backup sysres didn't work.
Backup backup sysres didn't work.
Backup backup backup sysres didn't work.
Backup backup backup backup sysres would IPL
enough to find and fix the problem.

   If you need something else, other methods
should be available.

[blackiron79]

--- In H390-MVS@..., "au1john" <au1john@...> wrote:
snip

Thanks for the links. I actually had this doc downloaded but didn't remember having it!

[blackiron79]

--- In H390-MVS@..., "somitcw" <somitcw@...> wrote:
>    Not all SVCs can be intercepted with all methods
> ( SVC Filtering, SVCTABLE update, ZAP, LPALIB load
> module replacement, LPALIB module replacement link,
> NUCLEUS relink, etc. ) but having a GETMAIN SVC
> that is impossible to intercept does sound far out.

I am not aware of any SVC that can't be intercepted by updating the SVC table or by subsystem SVC screening. But now that I think of it and after reading Gerhard's later comments, it would seem the problem was not that they couldn't hook the SVC but that they couldn't replace the entire GETMAIN service, which also includes a branch entry.

The methods of intercepting SVCs I use are dynamic, our products are installed after the system comes up. If you want to write OS modifications so the OS comes up with your replacement services ready to go, then these 2 methods I mentioned aren't going to be the answer. They need a more static method like some of the ones you mentioned.

I'm responding to your reply, somitcw and also trying to address Gerhards comments in this same response, because they're related.

I think you should be able to work backwards from XA doc if you can find it, to understand the GETMAIN branch entry interface. I don't think there are too many surprises, only simple things like using R4 (what were they thinking!) as the address of the owning task, etc. for certain subpools.

Is the subsystem interface available in 3.8 and does it allow you to intercept broadcasts for EOT and EOM? If so maybe this is the answer to Gerhard's comment (which I didn't completely understand) about wanting to get control at those events.

Jack