home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
kermit.columbia.edu
/
kermit.columbia.edu.tar
/
kermit.columbia.edu
/
archives
/
deleteme.zip
/
ikrker.bwr
< prev
next >
Wrap
Text File
|
1993-10-21
|
10KB
|
190 lines
ROSCOE Kermit 4.3.x "Beware File" - Known Bugs, Problems, Limitations.
As of: 14:12pm Thursday, 1993 September 30
In order to install ROSCOE Kermit, it is first necessary to have all the
materials for TSO Kermit, specifically, the files with prefixes of IKT
and IK0 (I K zero). This file (IKRKER.BWR) and its companion file
IKRKER.UPD are supplements to the corresponding TSO Kermit files
IKTKER.BWR and IKTKER.UPD. Thus, ROSCOE Kermit is just a variant of TSO
Kermit (for the present, anyway) with a few extra updates to take care
of special problems in the ROSCOE/ETSO environment. In particular, it
still uses "Kermit-TSO>" as the prompt and announces itself to other
Kermits as "TSO" (via the attribute mechanism, but this is about to
change). At some time in the future, there will be more extensive
updates which take advantage of the features of ROSCOE, and users are
encouraged to consider adding support for ROSCOE-specific files. ROSCOE
Kermit has been reported to work successfully under both V5.6 and V5.7
of ROSCOE, but no reports have specifically mentioned any other
versions.
For the uninitiate, here is a brief description of ROSCOE:
ROSCOE is a TSO substitute created around 1980 and now marketed by CA.
It is a multi-user address space with editing and job-submitting
facilities and a CLIST-like language called RPF (ROSCOE Programming
Facility). One of the components of ROSCOE is the Extended Time Share
Option (ETSO), which allows many programs written for TSO environments
to function under ROSCOE. However, there are often problems, due
mainly to the fact that everyone is in the same address space. This
can cause memory to run short or applications not to work at all. The
best thing about ROSCOE is its file structure, which compresses data
automatically and keeps DASD VTOC's from getting cluttered. Each user
is allotted a certain amount of space and is assigned a library prefix
for the members he or she creates. Files for all users are stored in
(typically) 5 datasets located on different packs.
This file contains (A) instructions for applying Kermit updates, (B) a
list of unresolved problems, (C) a list of problems with work-arounds or
optional fixes, and (D) a list of resolved problems.
NOTE: entries will sometimes be entered under (D) even though the needed
updates have not been released in IKRKER.UPD. Please check the date of
the UPD file to see if it has been changed before requesting a new copy.
Bug reports should be sent to John Chandler <JCHBN@CUVMB.CC.COLUMBIA.EDU>.
Some of the reports received to date have been included more-or-less
verbatim in parts (A)-(D) below and have been designated with the
initials of the correspondent: SC = Stuart Campbell, JL = John Lundberg,
DS = Dave Saunders. Dave Saunders also provided the summary of ROSCOE
above.
Kermit-370 carries an "edit" number in addition to the version and
release numbers. The three numbers together are printed out in the
assembly listing and displayed by Kermit at startup time. Be sure to
include all three numbers (e.g., 4.3.0) along with any error reports.
A) Update instructions .................................................
Refer to the instructions in IKTKER.BWR and IKTKER.INS. The only change
in procedure for ROSCOE is to include the ROSCOE-specific updates in the
KRMUPD step. This can be done quite easily by concatenating these
updates to the SYSIN DD declaration as a second data set for input. It
is expected that TSO Kermit will undergo upgrades from time to time
without necessitating any change to the ROSCOE-specific information in
this file and IKRKER.UPD. You should therefore keep track of new
developments under both IKT and IKR.
---------------- 91/1/3 --- DS -----------------------------------------
To run properly in a ROSCOE environment, Kermit must be defined in the
ETSO EPL (eligible program list):
KERMIT 3000 1024 1024 D CP KERMIT PROGRAM
These numbers (CPU time slices and memory allocation limits) can be
increased or decreased as sites see fit. The column position of the
numbers is important for entries in the EPL. Also, Kermit has to be in
a ROSCOE ETSOPGMS DD concatenation.
------------------------------------------------------------------------
B) Unresolved problems/warnings ........................................
Refer to IKTKER.BWR first for such messages.
---------------- 88/10/20 --- JL ---------------------------------------
TSO Kermit running under ROSCOE/ETSO seems to work for us until we do a
micro-MVS transfer without specifying a receive file name, whereupon
ROSCOE dies. There are problems in trying to create a host data set
name that violates ACF2's rule.
---------------- 91/1/3 --- DS -----------------------------------------
The major limitation is the need to use only MVS files. There is no
support for ROSCOE library members (but see below).
---------------- 91/1/11 -----------------------------------------------
The MVS "userid" of any ROSCOE address space, and, thus, the PREFIX
initially used by Kermit is "ROSCOE"; it is not your ROSCOE library
prefix. Be careful, since having an invalid prefix for MVS disk I/O can
lead to disaster. For example, the Kermit DEBUG subcommand would use
the dataset ROSCOE.KER.LOG by default, no matter how many Kermit users
there may be at one time. It is a good idea to reset the PREFIX used in
Kermit to something else. Also, it may be necessary to SET FILE UNIT to
an appropriate group name. Note, however, that the means of typeout
interception always uses the dataset ROSCOE.KER.BUF, regardless of the
current Kermit PREFIX -- that is one reason why host commands are
disabled in ROSCOE Kermit.
---------------- 91/01/11 ----------------------------------------------
Regarding the problems with system commands (see below in part D), there
is some question whether *all* commands would cause the ROSCOE address
space to crash, but there is certainly a problem with "OPER". Further
research is necessary to determine whether it would be worthwhile to
allow a specific list of commands to be executed. Any contributions of
information on this topic will be welcome.
---------------- 91/2/7 --- SC -----------------------------------------
ROSCOE Kermit sometimes stops communicating properly with the micro. If
STOP doesn't help, and the host is accepting input, then pressing CLEAR
several times gets a "terminal I/O error", and Kermit is back in
control.
---------------- 91/2/7 --- SC -----------------------------------------
ROSCOE has an escape character that defaults to "..". Set this to
something else if you expect double periods in input from the remote
Kermit.
---------------- 91/3/29 -----------------------------------------------
ROSCOE Kermit has been tested only in fullscreen mode (i.e., through a
7171). There is no reason to suppose, however, that it would not work
in line mode.
------------------------------------------------------------------------
C) Problems with workarounds or optional fixes .........................
---------------- 90/8/10 -----------------------------------------------
On a system without TSO/E, the STFSMODE macro doesn't allow the
specification of NOEDIT=YES, which Kermit uses. However, if you don't
run VTAM, the STFSMODE macro isn't needed at all. Also, without TSO/E,
the TPUT macro won't have a NOEDIT option, and the NOEDIT=YES option in
STFSMODE is superfluous anyway. The following update should do the
trick:
./ * SC90223
./ R 05617000 $ 5617000 40 08/10/90 ->4.3<-
STFSMODE ON,INITIAL=YES Full-screen mode @SC90223
./ D 05669500 08/10/90 ->4.3<-
./ R 05676500 05677000 $ 5676600 200 08/10/90 ->4.3<-
LM 0,1,0(8) Load up registers for SVC @SC90223
Also, you may need to issue the Kermit subcommand SET SEND PARITY MARK
before attempting transfers.
---------------- 91/1/3 --- DS -----------------------------------------
Occasionally, when Kermit is initially called in ROSCOE, the call will
fail with a message complaining of an illegal MODESET SVC 107. However,
no adverse effects seem to come from this, so we trap this message in an
RPF in this manner:
TRAP
CALL KERMIT
IF S.TC EQ 99 THEN
CALL KERMIT
ENDIF
We've never seen it fail twice in a row.
------------------------------------------------------------------------
D) Resolved problems ...................................................
---------------- 91/1/3 --- DS -----------------------------------------
Certain "HOST" and "TSO" subcommands could cause the entire ROSCOE
address space to crash with an S0F9 ABEND. The TSO "HELP" command also
wouldn't function correctly. The draconian solution is to disable all
these subcommands so that users can not issue them. The fix: VS91003 in
IKRKER.UPD.
---------------- 91/01/04 ----------------------------------------------
Kermit would operate normally when only one user at a time was running
it, but a second user would trigger an ABEND. This was due to Kermit's
subtask for timing out in linemode. ROSCOE Kermit no longer does that.
The fix: SC91004 in IKRKER.UPD.
---------------- 92/01/10 ----------------------------------------------
ROSCOE Kermit is in the midst of an enhancement effort designed to add
support for the ROSCOE AWS and library members. It so happens that
Kermit can already upload to the AWS under the guise of a dataset named
'ROAWS'. The other functions are requiring extensive code modifications
(volunteer testers welcome!)
------------------------------------------------------------------------