TSO rawhide published 9/3/2014
generated 9/3/14, open
1. noticed the READY READY issue
in z/OS. This is a long standing issue but not one I have seen previously except
in legacy MVS with the tcp=1$ driver. What happens is that, now and then, after the
colored 'KICKS is shutting down' message is shown you don't get the TSO ready
prompt. When you press enter again you get two READY prompts (usually white, not
red in legacy MVS). This does not happen with tcp=2$ or with any CMS tcp...
- can't reliably make the problem appear, so difficult to tell if attempted
fixes work...
- maybe due to leaving 'strange' stuff on the screen (such as the SA orders
associated with the shutting down message when KICKS ends.
- fix? -9/3/2014 - changed KSSFPGM to do a clear screen after 2 second delay after
sending the colored shutdown mesage - so that the screen TSO sees will be
totally blank.
** not the problem, 'fix' had no effect, issue remains
- maybe due to not doing a read after colored letter write? Noted on z/OS that
the colored letters where not showing up!
- fix? -9/25/14 - changed KSSFPGM to do a 'check' after colored letters write.
** 10/4/2014 nope, not fixed - READY READY appeared again (in
legacy MVS).
- more info -- 2/16/2015 can now reliably generate issue (on legacy MVS).
Generating procedure does not
generate ready-ready on z/OS, but does display possibly related issue in that
the KB is not locked after clear. Also noticed that I need to use QWS to test
this, as vista tn3270 seems to have a bug this procedure exposes ('short read'
for enter aid following clear aid). Also confirmed my "expected" result is
correct by trying in real CICS, and confirmed that it works properly in all CMS
versions of KICKS (and, within driver limitations, with tcp=2$ in
legacy MVS). At this point I'm testing to see if ready-ready may
be related to KICKS being called, not a command processor.
reported 9/3/14, fixed 9/4/14
2. KEBRPGM does not compile in
z/OS
due to compile diagnostic complaining about kik-aids1 being moved to a pointer
type.
- this due to LOAD api (for KEBRHELP) generating code for 'move after' from
kik-aids1 used (off label) for entry/set to a pointer.
- fix: updated KIKPPCOB (and PPGCC) to use given arg directly.
- also, while I was at it, updated PPCOB, PPGCC, KIKPCP1$, and KEDF to separate
SET/ENTRY (now reporting both) and to make all LOAD args optional.
reported 9/3/14 - resolved 9/5/14 - not a problem
3. IEC161I 020-054 abend (re
SDBDATA) in TSO when trying to use source trace. (this is z/OS system) also "1st startbr failed, resp=84, resp2=50" message in kedf
- this is an 'out of memory' situation. What's the region (below the line) for
the TSO session?
- resolved - 2m region, KICKS doc says min 4m
reported 9/4/14, fixed 9/4/14
4. GCC test program LINKTST fails
due to 'testcob3' not abending!
- this due to testcob3 not detecting condition to do FCP access to file NOFILE,
thus not abending. This due to PPGCC generating bad code in the case of an api
call directly following an if (ie, no preceding '{'), When that occurred PPGCC
gened a {linenum} {api expansion}, meaning the api expansion would not be within
the if as intended.
- fix: updated PPGCC to add another set of brackets so it looks like {{linenum}
{api expansion}}
- also, while I was at it, updated PPGCC to turn (8) error for missing KIKEIB in
main routine def into warning (4).
reported 9/6/14 queued to 1.5.1 release
5 Some GOLD and PRO markup still
exists.
- 1.5.0 is a 'PRO' version, but since that's the only kind there is now the GOLD
& PRO markup, thou still recognized, should be removed and when finally gone
PPCOB and PPGCC should have their recognition of it removed as well.
CMS rawhide published 9/7/2014
generated 9/9/14, fixed 9/9/14
6 KUNLKED vrm wrong, so stamped
in vrm in KIKENTRY is wrong for anything using it Most everything really...
- this is not vrm that is checked, except by eyeball, so it's ok in the rawhide,
but needs fixing before another publish
- KUNLKED updated, fix will propagate next ADOALL...
TSO, source, and CMS rawhides published 9/22/2014
generated 9/23/14, closed 9/28/14, all items addressed
7 z/OS source install issues
NOTE - z/OS is not a "supported" build platform for KICKS,
but since some may want to casually play with the source on z/OS it seemed best to address
these issues by fixing them when easy, and by otherwise documenting them...
- source install does not support update to JCLLIB for z/OS fixed 9/25/14
- ADOALL jobs run out of order due to multiple initiators serving class A jobs.
Need to fix to run right, or run one at a time instead of using ADOALL.
- added notes to ADOALL fixed 9/25/14
- FB2VB, PARMINFO, PDSUPDTE & KIKALI do not compile various JCL issues
- parminfo - not using proc, using mvt COBOL compiler
not gonna fix, not used in z/OS closed 9/25/14
- pdsupdt$ - using ASMFCL proc, doesn't work in z/OS
not gonna fix, used only in install, closed 9/25/14
- fb2vb - fixed 9/24/14
- kikali
not really a 'source' issue, but exposed in source testing...
fixed 9/25/14
- MUR & TAC examples do not compile
- these should compile from .CB2 lib, not from .COB lib
added notes to ADOALL & KIKSAMP fixed 9/25/14
- ASRB (ASRA, AICA) routine does not compile. also SIP won't
link due to missing ASRB routine
- jcl issue
fixed 9/25/14
- COB glue routine does not compile.
- jcl issue
fixed 9/25/14 by making it like GCC glue
- KEDFPGM compile fails
not really a 'source' issue, but exposed in source testing...
- bad PPCOB code gen, LOAD api expansion exceeds col 71
- result of issue 2 fix??
fixed 9/25/14
- KSDBLOAD compile fails
not really a 'source' issue, but exposed in source testing...
- cob2 doesn't like redefining 01 levels under FD
fixed 9/25/14
reported 9/23/14, fixed 9/25/14
8 z/OS install does not fully
support userid/prefix/literal HLQ due to missing support in the 2nd level rexx
($$V1R5M0) in the xmi.
- rewrote rexx to provide arguments allowing HLQ to be overridden.
- while I was at it I also provided ability to override volser.
reported 9/26/14, fixed 9/26/14
9 license does not address
concern re LE/370 in distros leaving legacy (legacy MVS, VM/370) systems behind.
- added restriction to license against incompatible objects
reported 9/27/14, fixed 9/27/14
10 running test programs as part
of ADOALL in source install takes to long for no purpose.
- made the last ADOALL step (running tests) optional
reported 9/27/14, fixed 9/27/14
11 tempstor, intrapartition
queues, and source db not defined during CMS install. This due to not running
LOADSYS in POSTVSAM exec.
- added LOADSYS to POSTVSAM
TSO, source, and CMS rawhides published 9/28/2014
reported 9/30/14, fixed 9/30/14
12 KEDF lines shown as green in
User's Guide are always blue...
- some tn3270 clients enable 3270 base color mode, some don't. Most don't allow
a choice! This is not an issue normally, but KEDF screens are really busy and losing the
color separation of its screen areas when using a tn3270 client that does enable
base color mode is not helpful.
- added attribute set into DOTOP to force green
last minute change! 10/2/14, fixed 10/12/14
13 TSO version needs KICKS...SKIKLOAD
in addition to KICKSSYS...SKIKLOAD (like CMS version already has). This to simplify user table maintenance for
"system install" versions of KICKS.
- updated both clist & exec to create user libs if not already present, and to
incorporate into appropriate concatenations.
reported 10/3/14, fixed 10/3/14
14 ksgmlic fails on mod 3 terms.
This due to incorrect map select (switched mod 3 & mod 4).
- corrected map select code.
reported 10/7/14, fixed 10/9/14
15 source trace compiles failing
IEC020I 001-4. compile proc(s) jcl issue, possible mismatch of blocksizes in
concatenation.
- fixed source trace compile procs.
reported 10/9/14, fixed 10/9/14
16 source trace reporting source
date non-match when should report source not found. "i16" (used in startbr
call) was short but should be int.
- made i16 int instead of short
reported 10/10/14, fixed 10/10/14
17 'funny' chars in KEBRPGM
source code. Actually just a few "[" and "]" characters mistranslated (more ind$file mess ups) in a couple of comments, but fixed anyway...
reported 10/10/14 - resolved 10/13/14 - not a problem (but maybe worth a
comment in doc or KooKbooK someplace)
18. KEDF source trace PF4 "skip
subroutine" did not return.
- skip subroutine returns when the program gets to the next executable instruction,
but that may not
be the same as the next instruction the program executes! For example, consider
IF blah blah THEN
PERFORM SKIP-THIS
ELSE
PERFORM MARKS-THIS-AS-RETURN.
PERFORM EXPECTED-RETURN-HERE.
if you were at "PERFORM SKIP-THIS" and pressed PF4 you might expect to return
to "PERFORM EXPECTED-RETURN-HERE", but KEDF sets up the temporary
breakpoint at "PERFORM MARKS-THIS-AS-RETURN", which is not hit so no "return" is
apparent.
- working as designed!
- possible enhancement: PF16 (shift PF4) to do "set temp BP at current
location (or
cursor) and go".
reported 10/16/14, fixed 10/17/2014
19 when using source trace, and
no source is found, and a resp<>0 event occurs, the resp is not shown,
which is confusing since the only reason for the break is the bad resp! This due
to not calling api_resp before dobottom in those error cases in source_trace.
- added required api_resp calls.
generated 10/20/14, open
20 this is to document some
'rough edges' in the KEBR program..
- analyze-cursor sometime reports wrong when cursor on "**bottom**" or bottom
ruler line of hex display
- repeated find sometimes does not switch directions properly
- need KooKbooK recipe for CICS install ✓(see issue 27)
- online help could be improved to show *all* commands instead of just those new
or different from CEBR; also vget should note the vsam file name is as in the
FCT (ie, its DDN), not as in its DSN.
- 'checksum' partially implemented to detect queue changed since last displayed;
needs to be finished...
- find highlight could be improved. 'scroll to' highlighting a bit confusing
reported 10/20/14, fixed 10/23/2014
21 KEBR does not find short tdq/file
names in get/vget if an optional argument is present
- changed to move only up to first space to tdq/file name
reported 10/23/14, resolved 10/23/14 - not a problem
22 some dates in the issues list
are beyond the
"last updated date" at the bottom of the page
- "Last updated date"
refers to the Appendices document, but the issues list is a server side include
and can change without the Appendices document changing. Added 'issues last
updated' to bottom of the include...
feature request 10/30/14, request declined 11/11/14
24 show "source listing not found"
message only once per program per task instead of for every source statement.
- really only an issue for a program that was compiled with source trace, but
the source code database has since been reinitialized so the source is no longer
available. Options: recompile (with or without source trace), switch to API mode
(PF2), or just turn off
source trace (PF13).
- 4/22/16 - changed 'SOURCE LISTING NOT FOUND' msg suffixing ', PRESS PF2 FOR API MODE'.
TSO, source, and CMS final (?) published 11/7/2014
reported 11/9/14, resolved 11/9/14 - ** unpublished **
25 On TK4- system, where
KICKS "system install" done, tso source build ADOALL step ADO11 submitted jobs (LOADMUR, LOADTAC, LOADSDB, LOADTEMP, LOADINTR) fail due to RAKF issue, unable to delete vsam file(s).
- cause is that specific RAKF rules exist for these (system install use) files, and the default id
associated with intrdr job isn't allowed to delete. Maybe not a big deal, since
this step isn't actually necessary to the build; the files are already built in
the course of the normal KICKS install.
- solution: delete step 11 since it is not necessary anyway
observed 11/16/14, resolved 11/19/14 - ** unpublished **
26 Syncpoint exit may foul up
pseudo conversational tasks. Fixing this turned out to be one of those efforts
that kept turning up mostly unrelated minor issues that also got fixed along the
way, as noted.
- fixed KCP code calling exit to preserve calen & nextran
- also noted and fixed related issue in PCP setting nextran
- also noted and fixed tempstor & intrapartition queue 'empty startup' problem
- also upgraded SIP auxtrace to show more decoded trace entries
published 11/21/14
27 KooKbooK recipe 15, "Installing KEBR in real CICS"
generated 11/22/14 open
28 status of KooKbooK updates (1-14)
- 1 - Designing and implementing simple BMS maps - republished 11/19/2014
- 2 - Designing and implementing simple KICKS COBOL programs -
republished 11/21/2014
- 3 - Designing and implementing simple KICKS COBOL programs, part II - republished 11/22/2014
- 4 - Run Away Task! - republished 11/23/2014
- 5 - Designing and implementing simple
KICKS COBOL programs, part III
- republished 12/06/2014
- 6 - 251 project – Scripting 3270 apps with open source tools ( I ) -
republished 12/30/2014
- 7 - 251 project – Scripting 3270 apps with open source tools ( II ) - republished
1/16/2015
- 8 - 251 project – Resource Definition ( I ) - republished
1/17/2015
- 9 - 251 project – Resource Definition ( II ) - republished
1/18/2015
- 10 - 251 project – Workload & Final Project Wrap Up - republished
1/19/2015
- 11 - Designing and implementing simple KICKS GCCMVS programs - republished
2/3/2015
- 12 - Using Alternate 3270 screen sizes in COBOL and GCCMVS programs - republished
11/23/2015
- 13 - Using DB2 with KICKS (reformat only, content update
queued to
1.5.1 release) -
- 14 - VSAM in the VM/370
SixPack (1.2) system -
pending change 11/22/14 queued to
1.5.1 release
29 improve tempstor, intra queue
sharing
a. change FCP close so that, in CMS, rel/det happens after last open file is closed
b. close tempstor, intra on conversational read even if pct entry says files stay open
reported 11/26/14 open
30 PDSman.xls does not handle
member aliases. This utility is not distributed with KICKS, but it
is something I've posted to the KICKS web site so I will
fix this sooner or later. Since no KICKS distribution datasets
contain aliases it's not a high priority...
feature request 11/30/14 queued to
1.5.1 release
31 Claim is KICKS already
'does' IRC, it just doesn't name it. This request is to support, for
compatibility, specification of SYSID in api calls. Such calls would be
execution time comments, but since files/queues are (or can be) shared, the
result is the same as if 'real' IRC was happening. Request prompted by desire to
add IRC to KEBR for use in real CICS.
- looks like a good idea, may need further research and definition
pending change 12/16/14 queued to
1.5.1 release
32 improve trace, auxtrace & kedf
trace by adding eibline & eibprog to each trace entry (internal entry size will
go from 16 to 32). This may also allow backward lookup (how you got here) at a
source trace breakpoint...
pending change 12/31/14 open
33 update User's Guide Install
section for TK4- update 6, 7.
- also do more clean up of the whole install section...
observed 1/5/15, resolved 1/6/15 - ** unpublished **
34 noticed that VRM checking
apparently systematically (but accidentally!) removed from the suffixed
management modules, needs to get put back...
observed 2/6/15, fixed 8/8/2015 - ** unpublished **
35 directly returning current
comm-area does not work. Usual code is to move it to working storage, change in
working storage as needed, then return the working storage copy. But fails if
RETURN TRANSID(EIBTRNID) COMMAREA(INPUT-COMM) LENGTH(EIBCALEN)
- because INPUT-COMM is cleared before it's moved to itself...
- changed to clear remaining after move...
reported 2/6/15, resolved 2/10/15 - ** unpublished **
36 use of SYSALLDA in legacy MVS
install procedures causes install job failures when the allocation
occurs onto a pack without sufficient space.
- changed GETXMI & RCVKICKS (install instructions) and V1R5M0 (in BIGPDS
install library) jcl to
specify SYSDA instead of SYSALLDA.
last minute change! 2/15/15 - ** unpublished **
37 While working on issue
#1 (ready-ready) I made a temporary change to ACP to un-suppress normally
suppressed transaction dumps when intense tracing is active. Having decided this
is a worthwhile change I've decided to make it permanent.
last minute change! 5/8/15 - ** unpublished **
38 Minor license revision; Under
USE, authorized users to "agree to" as well as "not violate" terms.
reported 5/13/15, fixed 8/7/15 - ** unpublished **
39 screen shots in 'examples'
part of user's guide missing.
generated 5/17/15 open
40 TS QUEUE and QNAME arguments
(for DELETEQ, READQ, WRITEQ) need additional clarification in their API
description to state QUEUE expects 8 characters of name, and QNAME expects 16
characters of name. Either will be right padded with spaces to make 16
characters. Names in the queue itself are always 16 characters..
reported 12/18/15, fixed 12/18/15 - ** unpublished **
41 Install fails with "IKJ56221I
VOLUME ..." message trying to install KICKS (or KICKS source) onto specific
volume in z/os.
- happens when volume specified isn't part of the default volume group.
Fixed by adding 'UNIT(SYSALLDA)' to allocation in $$V1R5M0 rexx.
reported 5/4/16, fixed 6/26/16
42 license does not address
transfer.
- added 'non-transferable' to free license grant.
reported 5/14/16, fixed 6/26/16 - ** unpublished **
43 Cobol preprocessor with source
trace option generates bad code when using as a pre step for a DB2 program.
- 'exec sql' block must not have trace statements inserted!
feature request 7/1/16 queued to
1.5.1 release
44 Would be helpful if SPOOLOPEN
supported nocc/asa/mcc options for carriage control..
- nocc is what it's doing now, but some 'printables' (like KEBR help) have asa
control.
- looks like a good idea, may need further research and definition (re
additional options beyond those mentioned)
reported 8/14/16, fixed 8/20/16 - ** unpublished **
45 Compiling for 'no KEDF' does
not work in TSO, program still traps into KEDF when it's turned on. Works
in CMS.
- fix for issue 7 introduced new bug KIKCOBGX same as KIKCOBGL! Replaced 'undef
DOEDF' for KIKCOBGX compile (like in CMS).
- added 'with source trace' to KEDFPGM 'on/s' response while fixing this.
reported 3/21/17 open
46 CONDITION(s) in HANDLE
CONDITION (etal) should be documented in Users Guide, rather than just saying
they are the same as DFHRESP (which is itself undocumented in Users Guide).
feature request 4/1/17 queued to
1.5.1 release (or later)
47 Would be helpful if RETURN
TRANSID
supported IMMEDIATE option..
- helpful in a db2 environment.
- might be a good idea, needs further research, especially wrt INPUTMSG.
reported 7/30/17 open
48 0C1 abend in batch KICKS
following install of v8 TK4-.
- circumvention: add LDMOD S37X to tk4-.cnf to restore 37x opcodes...
- long term: need to change the way KICKS detects 31 bit capable/zos...
reported 1/25/20, fixed 1/25/20 - ** unpublished **
49 Protection Exception in
KIKCB2CL EXEC (zvm). Caused by lack of support for new format libraries in
KIKPPCOB.
- circumvention: dont use new format maclibs, so if KICKS EXEC creates some user
libs on A disk (as it will if you install KICKS elsewhere), move the supplied
oncs from there to your A disk so they will be in 'OLD' format...
- updated KIKMACLB subroutine to support new lib format (as well as old).
issues last updated 1/25/2020