Trailing-Edge
-
PDP-10 Archives
-
BB-D868C-BM
-
4-documentation/exec.tco
There are 37 other files named exec.tco in the archive. Click here to see a list.
TOPS20 Change Order Number 4.1882
Written by: PORCHER 31-Jan-78 15:51:34
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0 EXECPR EXECPR
Related TCO's:
Related SPR's:
Problem:
EXAMINE and DEPOSIT commands need to be more useful.
Diagnosis:
Solution: Make EXAMINE and DEPOSIT default addresses.
Defaulting will be done as follows:
- There will be no initial default
- EXAMINE will default to the last address EXAMINEd plus one
if the last operation was EXAMINE
- EXAMINE will default to the last address DEPOSITed
if the last operation was DEPOSIT
- DEPOSIT will default to the last address DEPOSITed plus one
if the last operation was DEPOSIT
- DEPOSIT will default to the last address EXAMINEd
if the last operation was EXAMINE
Thus, alternate EXAMINEs and DEPOSITs will always reference the same
address; while successive EXAMINEs or DEPOSITs will reference successive
locations.
TOPS20 Change Order Number 4.1883
Written by: PORCHER 31-Jan-78 16:04:47
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECPR EXEC1 EXECSU EXECGL
Related TCO's:
Related SPR's:
Problem: TAKE files don't echo.
Diagnosis: Feature not implemented.
Solution: Add feature.
New subcommand to the TAKE command:
@TAKE,
@@ECHO (COMMANDS FROM "TAKE" FILE)
@@
@
This subcommand causes all commands processed during a TAKE file
to be typed on the user's console terminal (or TAKE log file).
Each complete command is typed as it is processed, preceeded
by the prompt that would have been typed for the command.
Note that this mode stays enabled for all TAKE files until
it is disabled:
@TAKE,
@@NO ECHO (COMMANDS FROM "TAKE" FILE)
@@
@
Note that TAKE with no arguments but with subcommands only
affects the state of the subcommands.
TAKE with no arguments or subcommands terminates the current command
file level, as always.
TOPS20 Change Order Number 4.1963
Written by: HEMPHILL 31-Jul-78 11:07:12
Edit checked: YES Document: YES
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's: 11386
Problem:
DOCUMENTATION SAYS THAT TMOUNT GIVES:
?OPERATOR NOT IN ATTENDANCE (OR SOME SUCH MESSAGE)
IF THE OPERATOR IS NOT IN ATTENDANCE. IN ACTUALITY, TMOUNT
IGNORES THE OPERATOR-IN-ATTENDANCE FLAG.
Diagnosis:
NO CODE TO CHECK THE OPERATOR-IN-ATTENDANCE FLAG.
Solution:
ADD CODE TO CHECK THE FLAG. IN ADDITION, CHANGE MESSAGE TO A WARNING:
%OPERATOR NOT IN ATTENDANCE (SAME AS SMOUNT'S WARNING)
THEN, GO AHEAD WITH THE TMOUNT REQUEST. CHANGE THE MESSAGE FOR TMOUNT
FROM [OPERATOR NOTIFIED] TO [WAITING FOR TAPE TO BE MOUNTED].
TOPS20 Change Order Number 4.1976
Written by: OSMAN 9-Aug-78 17:09:53
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2
Related TCO's:
Related SPR's:
Problem: "?" ON EXEC "TYPE" COMMAND CLAIMS
THAT YOU MIGHT WANT TO GIVE SUBCOMMANDS, WHEN IN FACT THERE ARE NONE.
Diagnosis: "TYPE" WAS CALLING CFN1 WITH THE WRONG BITS SET
Solution: INSTEAD OF CALLING $INFGX TO READ THE FILESPECS, CALL CFN1
DIRECTLY WITH THE APPROPRIATE BITTIES SET.
TOPS20 Change Order Number 4.1977
Written by: OSMAN 10-Aug-78 10:09:20
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1 EXECSU
Related TCO's:
Related SPR's:
Problem: UNDELETE COMMAND IN EXEC ADVERTISES
SUBCOMMANDS (IF YOU TYPE "?" AT STRATEGIC SPOT), WHEN IN FACT THERE
ARE NONE.
Diagnosis: NO SPECFN ENTRY WHICH SAYS "FILESPEC LIST IS LAST ON LINE
AND NO SUBCOMMANDS EXIST.
Solution: CHANGE OLD B17 DEFINITION TO MEAN "NO SUBCOMMADS". CALL
SPECFN FROM .DELET ARMED WITH THIS BIT.
TOPS20 Change Order Number 4.1978
Written by: OSMAN 10-Aug-78 14:32:49
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU EXECCS
Related TCO's:
Related SPR's:
Problem: EXEC COMMANDS LIKE "DIR/L" OR "COMP A=B" USE MUCH COMPUTES,
AND THEN BOMB OUT WITH "?TOO MANY JFNS IN COMMAND"
Diagnosis: EXEC LOOPS ON THE NULL STRING BEFORE THE "/" OR "=",
CONTINUALLY GETTING A SUCCESSFUL RETURN FROM GTJFN (THROUGH COMND).
Solution: REQUIRE COMMAS BETWEEN FILESPECS IN DIRECTORY COMMAND.
BE MORE STRICT ABOUT COMPILE COMMAND SYNTAX (DON'T ALLOW MORE THAN
TWO FILESPECS IN A ROW)
TOPS20 Change Order Number 4.1979
Written by: OSMAN 11-Aug-78 10:32:40
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: IN EXEC, TYPING "SYS D.SCHEIFLER" PRODUCES WRONG
RESULT. IT SHOULD SHOW INFO ABOUT USER D.SCHEIFLER, BUT DOESN'T.
Diagnosis: EXEC SEES "D" AS KEYWORD "DIRECTORY", "." AS MEANING
SELF, AND "SCHEIFLER" AS ANOTHER ARG.
Solution: CHECK TERMINATOR OF KEYWORD, IN THIS CASE "DIRECTORY".
IF IT'S ".", REINTERPRET INPUT WITHOUT KEYWORD AS A POSSIBILITY.
THIS WILL CAUSE "D.SCHEIFLER" TO BE SEEN AS A USER NAME.
NOTE THAT THE EXEC ALREADY DOES THIS TYPE OF TRICKERY FOR THE CASE
OF:
@SYS:MACRO
THE EXEC FIRST SEES THE "SYS" AS SYSTAT, AND THEN SAYS, "OH, IT'S
TERMINATED WITH ":" SO MUST NOT REALLY BE SYSTAT, BUT MUST BE THE
BEGINNING OF A FILESPEC!"
TOPS20 Change Order Number 4.1981
Written by: OSMAN 12-Aug-78 13:17:27
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2 EXECDE
Related TCO's:
Related SPR's:
Problem: COPY COMMAND IS TOO SLOW
Diagnosis: IT DOES TWO PMAPS FOR EVERY FILE PAGE COPIED. IT
ALSO PREFAULTS PAGES INTO A SEPARATE 8-PAGE BUFFER
Solution: DO MULTIPLE-PAGES AT A TIME. ALSO, DON'T USE A SEPARATE
PREFAULT BUFFER.
TOPS20 Change Order Number 4.1982
Written by: OSMAN 12-Aug-78 16:21:52
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EVERYTHING
Related TCO's:
Related SPR's:
Problem: COPY COMMAND IN EXEC BOMBS OUT IF PROBLEM OPENING INPUT FILE
Diagnosis: ROUTINE THAT INTERPRETS ERROR DOESN'T RETURN
Solution: IF INPUT FILE OPEN FAILS, PRINT MESSAGE AND GO ON TO NEXT
FILE IN GROUP
TOPS20 Change Order Number 4.1988
Written by: OSMAN 21-Aug-78 20:04:38
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: IN EXEC, THIS HAPPENS:
@DEFINE FOO: MUMBLE?
?INVALID HELP POINTER
Diagnosis: EXEC NEGLECTS TO SET UP HELP TEXT CORRECTLY
Solution: DO SO
TOPS20 Change Order Number 4.1995
Written by: OSMAN 29-Aug-78 11:40:30
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECQU
Related TCO's:
Related SPR's:
Problem: CAN'T SAY "CANCEL BATCH /SEQ:123" WITHOUT PRECEDING
/SEQ WITH A SPECIFIC JOBNAME
Diagnosis: EXEC WAS NOT HANDLING NON-SPECIFICATION OF JOBNAME
CORRECTLY
Solution: IF NO JOBNAME SPECIFIED, ASSUME "*"
TOPS20 Change Order Number 4.1996
Written by: OSMAN 29-Aug-78 11:41:30
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECQU
Related TCO's:
Related SPR's:
Problem: LAST TCO
Diagnosis:
Solution: SAME FOR MODIFY COMMAND
TOPS20 Change Order Number 4.2001
Written by: DBELL 30-Aug-78 23:51:22
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: ^ESENDS DO NOT WORK TO PARTICULAR TTYS
Diagnosis: EXEC IS HANDING TTMSG TTY NUMBER, NOT TTY DESIGNATOR
Solution: IF SENDING TO PARTICULAR TTY, ADD IN TTY DESIGNATOR.
TOPS20 Change Order Number 4.2006
Written by: OSMAN 14-Sep-78 15:36:32
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: INFO MONITOR prints "core" instead of "memory"
Diagnosis:
Solution: print "memory"
TOPS20 Change Order Number 4.2009
Written by: OSMAN 15-Sep-78 11:54:20
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: EXEC PRINTS AN EXTRA "PASSWORD)" IN HALF-DUPLEX MODE
Diagnosis: HISTORIC BUG
Solution: REMOVE EXTRA STRING
TOPS20 Change Order Number 4.2010
Written by: OSMAN 15-Sep-78 19:25:03
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: When commands like "directory A,B,C,D"
are given, and one of the filespecs is incorrect, exec doesn't always
print the correct name in the error message.
Diagnosis: counters and pointers not handled properly
Solution: If filespec parse fails, use parse-only. Use result of that for
error message.
TOPS20 Change Order Number 4.2012
Written by: OSMAN 17-Sep-78 19:54:59
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: if you define snark: as foo: and then say delete foo.*,
exec prints strange error message
Diagnosis: Although the GTJFN on FOO.* succeeds, the JFNS reveals SNARK:
as the device, which RCDIR then can't find because FOO: doesn't exist.
(we were connected to real SNARK:!). Exec was calling an error routine
which assumed AC1 contained an error code. In this case, however, RCDIR
didn't fail, it merely had a no-match.
Solution: Print real error message if RCDIR fails
This type of problem probably happens all over the place. That is, define
a logical name that's the same as your connected structure, such that the
definition points to a nonx structure. Then reference a file on DSK: such
that exec or other prog ends up trying to take the jfn, do a JFNS, and then
tries to feed the structure string to another jsys.
TOPS20 Change Order Number 4.2015
Written by: OSMAN 19-Sep-78 17:08:08
Edit checked: YES Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECQU
Related TCO's:
Related SPR's:
Problem: YOU GET A FUNNY ERROR MESSAGE ON "PRINT TTY:"
Diagnosis: EXEC CALLS WRONG ERROR ROUTINE WHEN GTFDB FAILS
Solution:
THE FRONT-END SHOULD INHIBIT THE DM-11/BB'S WHENEVER THE KL10 GETS
RELOADED. THIS WILL PREVENT THE AUTO-BAUD TASK FROM RUNNING AND
PRESENT A MORE CONSISTENT APPEARANCE TO A REMOTE USER AS THE MODEMS
WILL NOT ANSWER UNTIL DIRECTED TO DO SO BY THE KL10 MONITOR ONCE
IT HAS BEEN RELOADED AND HAS STARTED PRIMARY PROTOCOL.
TOPS20 Change Order Number 4.2024
Written by: OSMAN 26-Sep-78 13:29:57
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: PASSWORDS ECHO IF USER NAME TYPED INCORRECTLY DURING LOGIN
Diagnosis: EXEC TYPES ERROR BEFORE READING PASSWORD
Solution: READ WHOLE LOGIN BEFORE GIVING ERROR
TOPS20 Change Order Number 4.2037
Written by: OSMAN 7-Oct-78 00:43:20
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: In the exec, the SET PAGE-ACCESS command is too verbose about
reporting failures. Also, its failures were coming out as warnings instead
of real errors.
Diagnosis: I didn't feel like making the code smart before. Tonight I'm
ambitious.
Solution: Be clever and print only one error per block of consecutive pages
that fail for same reason instead of error per page. Also, print as real
error instead of warning. This was previously a warning merely because there
didn't used to be an internal mechanism in the exec to print an error and return
to the caller. Now there is so we can print real errors without bombing
out the command before all requested pages have been processed.
TOPS20 Change Order Number 4.2039
Written by: OSMAN 10-Oct-78 15:45:06
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: If user specifies more than one file in a TYPE command, the exec
is supposed to name each file as it types it out. The command
"TYPE A,B" fails to announce each file however.
Diagnosis: this worked in r2, and seems to have been broken in r3 and since.
The CFN1 routine in the exec was changed dramatically in r3, and the bug
was introduced there.
Solution: Light the flag saying a file group has been input (GROUPF), whenever
a subsequent filespec is guaranteed (such as user typing a comma after previous
filespec, in a situation where subcommands are impossible).
TOPS20 Change Order Number 4.2058
Written by: HEMPHILL 21-Oct-78 19:41:39
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECPR
Related TCO's:
Related SPR's: 11714
Problem:
When an EXEC is built without symbols, the last page is sometimes
missing in the .EXE file.
Diagnosis:
EXEC expects the last required address to be XEND. However, LINK
loads in SYS:MACREL.REL after this address, and PAT.. is placed
after this. If this additional code and storage crosses a page
boundary, it will not be saved.
Solution:
Make EXEC use one less than the contents of .JOBSY as the last
required address. That location points to the first symbol in
the symbol table.
TOPS20 Change Order Number 4.2059
Written by: HEMPHILL 22-Oct-78 07:39:10
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's: 11859
Problem:
No warning is given if a user tries to TALK to himself. The TLINK gives
a skip return, but no link is made.
Diagnosis:
If the user gets a warning when he tries to ADVISE himself, then he should get
a warning when he TALKs to himself.
Solution:
Add a test and an error message.
TOPS20 Change Order Number 4.2077
Written by: OSMAN 4-Nov-78 01:13:52
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECPR
Related TCO's:
Related SPR's:
Problem: If you are doing alot of fancy things with exec such as "FORK n"
commands, the exec gets confused and a "DDT" command can cause
?Illegal instruction 0,,0 at 770000
Diagnosis: Exec has cell DDTFLG which it sets when DDT gets loaded, such that
subsequent DDT command can be efficient instead of figuring out whether to
load DDT again. After FORK command, exec gets fooled by DDTFLG into thinking
that DDt is already loaded for other fork!
Solution: clear DDTFLG whenever FORK command is given
TOPS20 Change Order Number 4.2082
Written by: OSMAN 7-Nov-78 14:42:50
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: On heavily loaded system, when exec starts on new job, it can type
?JSYS error at nnn
?No space available for a clock
and then go into an infinite loop of typing an error message
Diagnosis: this TCO only covers one problem. That is, at ERRFIN, there are
some PUSH'S and POP's, and the POP's win by one!!!!! Since the exec doesn't
reset P after an error, this cause a flushdown plunderflow.
Solution: Remove the PUSH's and POP's completely. (Noone was expecting the
ac's to be preserved there anyway)
TOPS20 Change Order Number 4.2086
Written by: OSMAN 9-Nov-78 14:43:30
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC4
Related TCO's:
Related SPR's:
Problem: ON "INFO DIR", EXEC CAN SAY "DIRECTORY GROUP BUFFER OVERFLOW" WHEN
NOT SO.
Diagnosis: EXEC WAS CHECKING TO SEE IF LAST WORD IN BUFFER WAS 0 AFTER GTDIR,
WHICH IS INVALID, SINCE THE WORD ISN'T INITIALLY 0 BEFORE THE GTDIR.
(THIS PROBLEM DIDN'T EXIST WHEN THE GROUP BUFFER WAS IN GLOBAL INSTEAD OF
LOCAL STORAGE IN EARLIER RELEASES, IN WHICH ALL STORAGE IS ZEROED ON EVERY
COMMAND)
Solution: CHANGE THE CHECK TO BE TO SEE IF THE COUNT OF GROUPS RETURNED BY
GTDIR IS MAXIMUM, IN WHICH CASE PRINT THE OVERFLOW MESSAGE.
NOTE THAT GTDIR IS DEFICIENT, AS IT HAS NO WAY OF TELLING THE CALLER THAT
THE ALLOTTED GROUP SPACE ISN'T LARGE ENOUGH FOR ALL GROUPS. HENCE FOR
DIRECTORIES WITH EXACTLY THE MAGIC NUMBER OF GROUPS, THE EXEC FALSELY SAYS
"GROUP OVERFLOW"
TOPS20 Change Order Number 4.2087
Written by: OSMAN 10-Nov-78 10:06:47
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECPR
Related TCO's:
Related SPR's:
Problem: If you have a COMAND.CMD that sets up some defaults for you, and you
run an exec for the purpose of patching it, and put the patched one up on
SYSTEM:, everyone gets your defaults! That is, you patch it like this:
@RUN SYSTEM:EXEC
@ENA
$^EDDT
patch patch patch patch...
CINITF/ -1 0
^Z
@SAVE SYSTEM:EXEC
Diagnosis: exec wasn't clearing defaults upon startup
Solution: Move the default database to the section of memory that the exec
clears upon first startup. This way, whenever the exec is started with CINITF
set to 0, it will clear all defaults. Hence it's O.K. to save a patched exec.
TOPS20 Change Order Number 4.2092
Written by: DBELL 19-Nov-78 17:52:24
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem:
When a user gets a message from somebody, he can't reply to it easily
because the user could be logged in on multiple terminals, or could
have logged out quickly.
Diagnosis:
All you get in the send message is the user name.
Solution:
Also give the terminal number in the message.
TOPS20 Change Order Number 4.2110
Written by: OSMAN 6-Dec-89 10:43:16
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU EXECPR EXECGL
Related TCO's:
Related SPR's:
Problem: In exec, if you use BUILD or ^ECREATE, and try the KILL subcommand,
and don't confirm properly, you're supposed to be back at subcommand
level. Observation shows that you cannot succeed in typing any more subcommands
at this point.
Diagnosis: Exec wasn't properly saving P away at beginning of subcommands, to
be restored after an error during a subcommand. The problem only
shows up if a subcommand causes a prompt, as with [CONFIRM] under KILL.
Solution: Recreate cell called .P which P gets stored in at beginning of
subcommands, and which error handler copies to P after an error if it's
non-zero.
TOPS20 Change Order Number 4.2111
Written by: OSMAN 7-Dec-78 11:18:57
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU execqu EXECGL
Related TCO's:
Related SPR's:
Problem: In exec SUBMIT command, one can't say /TIME:25:0:0, for instance.
Diagnosis: Exec uses monitor's dat-and-time JSYS's (via COMND) to read
the time. These JSYS's restrict input to 24 hours, because they assume
a time of day is being read, rather than an amount of time.
Solution: Make exec not use that method. Instead, exec will call
GETAMT, a local routine it already uses for reading amounts of time
in other places (like /AFTER:TODAY+45:0:0)
TOPS20 Change Order Number 4.2125
Written by: OSMAN 20-Dec-78 10:51:08
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: The exec command "COMP FOO + BAR" causes
?Internal error at nnn
Diagnosis: Parser of COMPIL-class commands doesn't expect spaces after "+".
Solution: Enhance CMPRES routine to remove spaces after pluses. (It already
removes lots of other spaces, like ones before slashes)
TOPS20 Change Order Number 4.2130
Written by: OSMAN 21-Dec-78 14:06:48
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: "LOAD @FOO" typed to exec causes "?Not confirmed".
Diagnosis: This recently broke when a COMND fix caused COMND to no
longer disallow "@" to be typed in lieu of guidewords. Exec turns
on CM%XIF for LOAD-class commands, but not for their guidewords!
Solution: In exec, before parsing "(FROM)" as in "LOAD (FROM)", turn
on CM%XIF. (Exec was already turning on CM%XIF for parsing the
rest of the command)
TOPS20 Change Order Number 4.2149
Written by: OSMAN 4-Jan-79 19:46:05
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: The exec command LOAD FOO %"text" fails to call the
compiler.
Diagnosis: The exec was confusing the space after "FOO" and thinking
an object filespec was forthcoming
Solution: Enhance CMPRES in EXECCS to throw away spaces in front of
percents.
TOPS20 Change Order Number 4.2159
Written by: OSMAN 12-Jan-79 17:08:06
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: YES
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: 1) Certain errors such as "INFO BLECCH" just type
"?" without an error message.
2) Certain errors such as "PRINT /YECCH" print two error messages,
i.e. "?" at the beginning of two consecutive lines.
Diagnosis: CERR was just typing "?". $CMERR (invoked by CMERRX) was
typing two lines
Solution: Make CERR do a CMERRX. Fix $CMERR to print single line
error message.
What this heiroglyph means for documentation:
1) Any manuals which document that the exec prints a lone "?"
should be changed to say the exec prints an appropriate error
diagnostic.
2) The double-line error messages in which both lines started with
"?" now print a single error message line.
TOPS20 Change Order Number 4.2167
Written by: OSMAN 17-Jan-79 13:48:06
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: Exec scenarios like
@SET DEFAULT SUBMIT /TIME:1:2:3
@SET DEFAULT SUBMIT /TIME:4:5:6
@INFO DEFAULT SUBMIT
produce
SET DEFAULT SUBMIT /TIME:1:2:3 /TIME:4:5:6
Diagnosis: exec neglects to cull out old specifications
Solution: smarten exec to find "duplicate" and remove it
TOPS20 Change Order Number 4.2172
Written by: OSMAN 25-Jan-79 13:32:15
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: INFO MEM takes too long
Diagnosis: Three factors:
1) The RPACS JSYS in the monitor is now slower due to extended
addressing support.
2) The exec does an extra preloop of RPACS's to count number of
pages total so that total can be printed before map.
3) In the map printer routine, the exec seems to do RMAP and RPACS
per existent page. Only RMAP is needed.
Solution: 1) Buffer the output in a string so that the same JSYS that
gets the page info can be used to count how many pages.
2) Do only RMAP per page, not RPACS.
TOPS20 Change Order Number 4.2183
Written by: HEMPHILL 2-Feb-79 15:32:06
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2
Related TCO's:
Related SPR's: 11548
Problem:
The BYTE (SIZE) subcommand of COPY is invisible.
Diagnosis:
It was made invisible for no apparent reason.
Solution:
Make it visible. (Clear the CM%INV bit in the keyword table.)
TOPS20 Change Order Number 4.2184
Written by: OSMAN 8-Feb-79 13:30:12
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: THE COMMAND "^E SET NO RUN-TIME GUARANTEE" DOESN'T WORK
Diagnosis: WRONG AC BEING REFERENCED. I probably broke it several years
ago when I put COMND in the exec
Solution: fix it
TOPS20 Change Order Number 4.2189
Written by: OSMAN 14-Feb-79 13:53:15
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: In exec, if user tries to say null password by just typing
carriage return in response to
PASSWORD:
he just gets reprompted. Instead, exec would like to assume null password
in that case.
Diagnosis: COMND merely reprompts by default if user types return at
beginning of line.
Solution: Do .CMCFM at beginning of line, which COMND regards as a special
case. IF .CMCFM suckseeds, assume use is giving null password.
TOPS20 Change Order Number 4.2190
Written by: HEMPHILL 15-Feb-79 15:51:08
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's: 11690
Problem:
If you type "RENAME FOO.FOO <DBELL>" and file FOO.FOO doesn't exist, EXEC
gives ?FILE NAME NOT SPECIFIED
Diagnosis:
If FOO.FOO doesn't exist then MFOUT uses no default filename to parse
the output filespec. "<DBELL>" does not contain a filename, hence the
error message. The message doesn't occur if you have more than one
input filespec, because then MFOUT defaults the file and extension to
*.*
Solution:
If there is only one input filespec, and an error of the file not found
variety occurred, allow MFOUT to finish the parse by defaulting *.*
for filename and extension.
TOPS20 Change Order Number 4.2191
Written by: HEMPHILL 16-Feb-79 16:17:14
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECPR
Related TCO's:
Related SPR's:
Problem:
When executing a DDT command, if the program to be debugged has no symbol
table, EXEC chooses to merge in SDDT. Unfortunately, if the program is
running with the compatability package, SDDT overlaps part of PA1050.
Diagnosis:
You can't use both SDDT and PA1050 at the same time.
Solution:
Make EXEC merge in UDDT if PA1050 is present.
TOPS20 Change Order Number 4.2195
Written by: OSMAN 21-Feb-79 16:36:12
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: exec
Related TCO's:
Related SPR's:
Problem:
When debugging an exec problem which causes the error
?Illegal memory write internal trap
using the ^EEDDT feature of the exec loses.
Diagnosis: The ^EEDDT command write-enables the exec. Hence the bug
being chased is more difficult to find!
Solution: Change ^EEDDT to not write-enable the exec. The write-enabling
used to be necessary in order to allow breakpoints to be put in. However,
the new $W command in DDT makes this unnecessary.
TOPS20 Change Order Number 4.2214
Written by: OSMAN 14-Mar-79 15:59:47
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU EXEC3
Related TCO's:
Related SPR's:
Problem: Exec allows
@DIR
@@SINCE MONDAY
which one would expect would mean all files created since last Monday,
but doesn't work.
Diagnosis: Exec was calling same routine as /AFTER:, which interprets
MONDAY as next Monday!
Solution: Call separate routine. Now, the exec will allow all of the
following after SINCE and BEFORE:
1) Specific date and time
2) Day of the week or the word TODAY, each of which may be optionally
followed by a "+" followed by an amount of time. For instance,
@@SINCE TODAY +4:0:0
means list all files created after 4 in the morning today.
3) "-" followed by amount of time. For instance,
@@SINCE -0:30:0
means all files created in last half hour.
TOPS20 Change Order Number 4.2220
Written by: OSMAN 21-Mar-79 09:50:02
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: INFO MEM command in exec keels over if page is mapped to a restricted
jfn.
Diagnosis: JFNS jsys isn't prepared for error
Solution: put in ERCAL JFNSIL after JFNS. That's how INFO FILS does it.
TOPS20 Change Order Number 4.2223
Written by: OSMAN 2-Apr-79 13:16:09
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: exec
Related TCO's:
Related SPR's:
Problem: Strange errors when operator types "^ESET TERMINAL 4000 SPEED ..."
Diagnosis: Exec doesn't check terminal number
Solution: Create string "TTYnnn:" and ask monitor if terminal exists.
TOPS20 Change Order Number 4.2237
Written by: OSMAN 20-Apr-79 10:38:44
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execmt
Related TCO's:
Related SPR's:
Problem: In ASSIGN command, an error saying "jsys error at ..." can
occur.
Diagnosis: lost due to editing mistake
Solution: If ASND fails, print monitor's error.
TOPS20 Change Order Number 4.2238
Written by: OSMAN 20-Apr-79 15:35:21
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: COMP FIOSUB.FOR,FSUBRS.MAC command loses in exec.
You get
?SCNCDR COMMA REQUIRED IN DIRECTORY 0 from FORTRA.
Diagnosis: Exec sets up lengths of tmpcore files to too long.
Length should not include header word.
Solution: Shorten lengths
TOPS20 Change Order Number 4.2258
Written by: OSMAN 24-May-79 14:30:36
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: tmount
Related TCO's:
Related SPR's:
Problem: TMOUNT defines a null logical name instead of correct one.
Diagnosis: TMOUNT fails to remember name specified.
Solution: Set NAMEP to correct name in .TMOUNT.
TOPS20 Change Order Number 4.2261
Written by: EKLUND 29-May-79 14:10:55
Edit checked: YES Document: NO
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2
Related TCO's:
Related SPR's: 12605
Problem: On an EXEC COPY command the user settable word (.fbusw) in the
file's FDB is not copied.
Diagnosis: No one ever considered this useful. It now appears that some
users are using this word for information which is used to determine the
kind of file it is (i.e., what version of COBOL should be used to compile
it).
Solution: Copy the user settable word if possible. The code is very simple
and is in EXEC2.
TOPS20 Change Order Number 4.2267
Written by: OSMAN 1-Jun-79 13:43:02
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: Sometimes, exec types "MAIL" in response to "COPY TTY:$".
Diagnosis: SET MAIL-WATCH causes MWATCH to be called to check for new mail.
MWATCH causes a jfn on MAIL.TXT to be stacked. The COPY command calls
.INFG with flags saying to default name to that of previous jfn, which
shouldn't exist, but MWATCH stacked one!.
Solution: Move the CALL RLJFNS, which cleans up jfns from previous command,
to after the CALL MWATCH.
TOPS20 Change Order Number 4.2268
Written by: OSMAN 1-Jun-79 14:49:48
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: OPR and EXEC both have ^ESET BIAS-CONTROL and ^ESET RETRIEVAL-WAIT
Diagnosis: They don't both need 'em
Solution: rip out of exec
TOPS20 Change Order Number 4.2270
Written by: OSMAN 4-Jun-79 10:25:59
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: If you ^C out of DIRECT [*]*.FOO for instance, a partially
created jfn lies around.
Diagnosis: I removed CLZFF from ^C loop to speed it up. Usually the
CLZFF isn't needed, since the exec stacks every jfn with JFNSTK, and
^C loop unwinds the stack. However, there are a few places, like
wildcard searching, that if user types ^C between GTJFN (in COMND!) and
the JFNSTK call, jfn can lie around.
Solution: Use new flag called CLZF to signal that CLZFF better be done if
^C. Set this flag before calling COMND (since @MUMBLE may be typed), and
clear CLZF after COMND comes back, if function successfully parsed was
not a filespec function. Clear the flag in JFNSTK too, after stacking the
jfn.
TOPS20 Change Order Number 4.2271
Written by: OSMAN 4-Jun-79 14:01:04
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: Exec leaves tape jfns around.
Diagnosis: Exec clobbers a jfn on the jfnstk at DI3, but should instead
leave it on the stack so it gets closed later.
Solution: yes
TOPS20 Change Order Number 4.2272
Written by: OSMAN 5-Jun-79 11:22:27
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: Bad error message on INFO ADDRESS-BREAK on 2020 after ^C
out of a program.
Diagnosis: Wrong syntax in instruction in exec
Solution: Put in the extra "%"!
TOPS20 Change Order Number 4.2274
Written by: OSMAN 6-Jun-79 09:59:02
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU EXEC3
Related TCO's:
Related SPR's:
Problem: Exec sometimes leaves jfn's around
Diagnosis: JFNSTK tries to remember what to close, but UNMAP wipes out the
info. Also, DIRFL / DI3 replaces tape jfn with another after rewinding, hence
the jfn stack doesn't remember all the jfn's that need closing.
Solution: Use GTBUFX / XBUFFS instead of GETBUF / BUFFF for stacking erroneous
jfn at CFNE1. That way, UNMAP won't hurt the jfn stack. Also, at DIRFL / DI3,
be more careful about leaving even the old jfn's stacked.
TOPS20 Change Order Number 4.2276
Written by: EKLUND 7-Jun-79 09:08:55
Edit checked: YES Document: NO
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's: 12372
Problem: EXEC gives poor error message for the class of commands like
"@DIR +".
Diagnosis: The code thought that the only way one could get there was
through having two file specs not separated by commas.
Solution: Change the error message to be more complete - "comma required
between filespecs or illegal character in command".
TOPS20 Change Order Number 4.2277
Written by: EKLUND 7-Jun-79 14:12:57
Edit checked: YES Document: YES
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC4
Related TCO's:
Related SPR's: 12539
Problem: With release 3 or later versions of the EXEC, one lost the
ability to change the repeat-login-messages bit for a user by using
a sub-command to the BUILD or ^ECREATE commands.
Diagnosis: The subcommand was inadvertently removed by some restructuring
of the code for version 3 of the EXEC.
Solution: Put the code back in to allow the setting and unsetting of the
capability via the BUILD command. The sub-command is: REPEAT-LOGIN-MESSAGES.
TOPS20 Change Order Number 4.2278
Written by: OSMAN 7-Jun-79 17:01:38
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: TIMES (and DATES OF) TAPE-WRITE
subcommand of DIRECTORY in exec doesn't work
Diagnosis: bit definitions screwed up
Solution: fix 'em
TOPS20 Change Order Number 4.2286
Written by: OSMAN 14-Jun-79 10:45:06
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: When RDMAIL is execute-only, which it better not be, exec
fails to print good message.
Diagnosis: TYPE used whenn ETYPE needed.
Solution: Use ETYPE
TOPS20 Change Order Number 4.2287
Written by: OSMAN 14-Jun-79 14:21:46
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECP
Related TCO's:
Related SPR's:
Problem: Bad error from exec when user tries to CONTINUE a program
before STARTing it
Diagnosis: SFORK recently has been fixed to o fail, so exec should trap it.
Solution: If user types CONTINUE command, and SFORK fails, such as because
program has never been started, exec will now print the error from the
SFORK.
TOPS20 Change Order Number 4.2293
Written by: OSMAN 18-Jun-79 14:35:15
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: CANCEL RETRIEVE works wierd
Diagnosis: It takes filespec, everything else takes job id.
Solution: Make it take job id. Rip out all other code, use regular code.
TOPS20 Change Order Number 4.2294
Written by: OSMAN 19-Jun-79 13:02:42
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: When class scheduling is on and there's no batch class, INFO SYSTEM
command in exec prints "batch class -1".
Diagnosis: -1 means none, not a real -1.
Solution: Make exec not say anything about batch class if it's -1.
TOPS20 Change Order Number 4.2295
Written by: OSMAN 19-Jun-79 14:18:46
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2
Related TCO's:
Related SPR's:
Problem: COPY /APPEND in exec doesn't do BYTE-SIZE 36 correctly
Diagnosis: TRO A, instead of TRO Z,
Solution: Make it TXO Z,
TOPS20 Change Order Number 4.2297
Written by: OSMAN 19-Jun-79 16:16:16
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execp-not-execpr
Related TCO's:
Related SPR's:
Problem: ATTEMPTing to SAVE and execute-only process gives wierd errors from
the exec.
Diagnosis: Extra PUSH at ILISSA
Solution: Remove the PUSH
TOPS20 Change Order Number 4.2300
Written by: OSMAN 20-Jun-79 16:20:29
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execmt
Related TCO's:
Related SPR's:
Problem: TMOUNT doesn't allow long names
Diagnosis: GETSXB called when GETSIX should be
Solution: CALL GETSIX
TOPS20 Change Order Number 4.2301
Written by: OSMAN 20-Jun-79 16:32:26
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: The following occurs:
@DELETE BLECCH, !BLECCH doesn't exist!!!
@@EXPUNGE (AFTER DELETING)
@@
%FILE NOT FOUND - BLECCH
garbage [No pages freed]
Diagnosis: The exec uses DELDIR cell for directory to be expunged.
If none of the specified files existed, DELDIR has "garbage"
in it.
Solution: Init DELDIR to 0, and don't call TYPFRE at all if DELDIR is 0.
TOPS20 Change Order Number 4.2302
Written by: OSMAN 21-Jun-79 11:13:14
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: The exec command LOAD A:FOO1+A:FOO2 fails if there is a current
.REL in A:.
Diagnosis: The MAKOBJ routine clears the ppn, which it shouldn't do.
Solution: Don't clear it!
TOPS20 Change Order Number 4.2303
Written by: OSMAN 21-Jun-79 11:30:35
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: Exec compiles on LOAD A:FOO even though .REL in connected directory
is current.
Diagnosis: Exec was only looking in source directory (A: in this case)
and didn't see the .REL in the connected directory.
Solution: Look in both connected and source directories, favoring connected
if .REL found there.
TOPS20 Change Order Number 4.2304
Written by: OSMAN 21-Jun-79 13:29:04
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: ^ESET TAPE-RECYCLE and ^ESET ARCHIVE-TAPE-RECYCLE periods
being removed from exec.
Diagnosis: They are in CONFIG and OPR, so let's not put them in exec too.
Solution: right!
TOPS20 Change Order Number 4.2310
Written by: OSMAN 26-Jun-79 08:54:08
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: When exec types
[Attached to TTY67, confirm]
and user types RUBOUT or CTRL/R etc., strange things happen.
Diagnosis: Exec prints that message as regular output rather than as a prompt.
Solution: Build that message in memory and display it as a standard prompt.
TOPS20 Change Order Number 4.2313
Written by: OSMAN 27-Jun-79 11:04:08
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: EXEC does two unnecessary SIN jsys's per file on DIRECTORY
command
Diagnosis: Exec does the SINs to put the string "UNKNOWN" into the buffer,
in case the call to get the last author and writer strings fails.
Solution: Don't do the SIN's unless the GFUST fails.
TOPS20 Change Order Number 4.2315
Written by: OSMAN 29-Jun-79 12:43:51
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: If you say RUN SYSTEM:EXEC and then try to list a directory
containing files with quoted characters, the quotes appear as ^V instead
of being invisible. This makes the filenames look ugly.
Diagnosis: Exec neglected to set up ^V with SFCOC on cold start.
It was, however, getting set up by the SYSLEV code handling system mail.
Solution: Call LTTYMD at cold start.
TOPS20 Change Order Number 4.2316
Written by: OSMAN 29-Jun-79 13:15:19
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: When doing long directories with exec, especially magtape
directories, the exec sits for a long time before typing anything.
Diagnosis: The exec buffers up the entire directory before printing,
since sorting may be required.
Solution: Don't buffer the directory unless sorting is required.
TOPS20 Change Order Number 4.2317
Written by: OSMAN 29-Jun-79 14:13:57
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0 EXECSU EXECDE
Related TCO's:
Related SPR's:
Problem: Sometimes ^T or INFO OUTPUT or INFO BATCH stops working
Diagnosis: Since ^C interrupt level was higher than ^T or IPCF, and
^C never DEBRKs, ^C out of IPCF or ^T would cause either to never cause
another interrupt. (This is because IPCF and ^T were at same level)
Solution: When clearing interrupt on ^C (ICLEAR), see if other levels already
in progress. If so, do CIS jsys to clear them. If CIS necessary, also do
IIC to catch small possibility of ^C out of IPCF interrupt routine.
TOPS20 Change Order Number 4.2325
Written by: OSMAN 13-Jul-79 11:00:42
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: Exec allows you to type /MODE on SUBMIT command.
Diagnosis: Exec table neglected to disallow /MODE on SUBMIT.
Solution: Fix it
TOPS20 Change Order Number 4.2326
Written by: OSMAN 13-Jul-79 14:45:53
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: Exec gets ILLEGAL MEMORY WRITE on INFO DIR PS:[*] when
enabled.
Diagnosis: The GETDRP routine tries to allocate a password block as large
as the supplied password. If none is supplied, however, GTDIR
overwrites free space header word.
Solution: In GETDRP, if no password pointer is supplied, allocate a block
of maximum size needed.
TOPS20 Change Order Number 4.2327
Written by: OSMAN 13-Jul-79 15:43:46
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU EXEC4
Related TCO's:
Related SPR's:
Problem: You get ?Exec free space exhausted
on INFO DIR PS:[*] when enabled.
Diagnosis: Each directory block takes up free space until...
Solution: Call new routine called RELDIR to release free space taken
up by directory.
TOPS20 Change Order Number 4.2328
Written by: R.ACE 15-Jul-79 08:22:44
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECMT
Related TCO's:
Related SPR's:
Problem: EXEC ITRAPS ON COPY TO MTA DEVICE
Diagnosis: .MORLI MTOPR FAILS FOR MTA DEVICES, AND THE LBLSKP ROUTINE
IN EXECMT DOESN'T TAKE THIS INTO ACCOUNT (IT ITRAPS DOING THIS MTOPR).
Solution: ADD ERJMP AFTER MTOPR IN LBLSKP ROUTINE TO TAKE +1 RETURN
INDICATING THAT MTA DEVICE IS UNLABELED.
TOPS20 Change Order Number 4.2331
Written by: OSMAN 17-Jul-79 09:47:37
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: COMPIL-class commands fail with comments in them, such as
@COMP FOO !This comment should be ignored!
Diagnosis: The exec parses the COMPIL-class commands twice, once with COMND
and once again with TEXTI and the exec's own special parsing routines.
COMND handles the comments fine. The exec trips on them.
Solution: In the CMPRES routine which already removes extraneous spaces, remove
comments too!
TOPS20 Change Order Number 4.2332
Written by: OSMAN 17-Jul-79 11:10:33
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: SUBMIT /AFTER:SATURDAY fails on Sunday. Also, @@BEFORE SUNDAY
(DIRECTORY command) fails on Tuesday.
Diagnosis: Sunday is day 0, and MOVNI C,-1(C) fails on this day only!
Solution: Put HRRE C,C after the MOVNI to fix Sunday bug.
TOPS20 Change Order Number 4.2334
Written by: OSMAN 18-Jul-79 10:12:47
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: The command COMP FOO.MAC with a trailing space causes a compilation
even though the .REL is current. This is particularly annoying with
situations such as
@COMP FOO.MAC !comment
which has a trailing space when the comment is stripped during parsing.
Diagnosis: None attempted. However, a preprocessor already removes spaces
in other places in the line for similar reasons.
Solution: Fix the preprocessor (CMPRES) to remove trailing spaces.
TOPS20 Change Order Number 4.2335
Written by: OSMAN 19-Jul-79 08:31:48
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2
Related TCO's:
Related SPR's:
Problem: SYSJOB.COMMANDS should go to SYSTEM:, not PS:[SYSTEM] as exec
currently does.
Diagnosis: There used to not be a SYSTEM: logical name. Exec was not updated.
This is usually not a problem, since SYSTEM: is usually PS:[SYSTEM].
Solution: Change references to PS:[SYSTEM] in ^ESPEAK command to SYSTEM:
TOPS20 Change Order Number 4.2338
Written by: OSMAN 24-Jul-79 13:29:56
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: The INFO MON command prints some load averages as .00 and some
as 0.00
Diagnosis: FLOUT prints .00 when the number isn't exactly 0.
Solution: Use the FL%ONE bit in FLOUT to force 0.00
TOPS20 Change Order Number 4.2347
Written by: OSMAN 26-Jul-79 13:07:10
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: When logging in to other systems with NVTHAK, you don't see
"You have a message" when you have one.
Diagnosis: Exec checks for PTY login and doesn't check for messages.
This is intending to check for batch.
Solution: Check for batch, not pty logins.
TOPS20 Change Order Number 4.2348
Written by: OSMAN 26-Jul-79 13:36:39
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: INFO DEFAULT SUBMIT command keels over if /TIME:30:0 is presiding.
Diagnosis: ODTNC hates times greater than 24 hours
Solution: Fire ODTNC! Employ manual labor instead.
TOPS20 Change Order Number 4.2351
Written by: OSMAN 26-Jul-79 17:07:58
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: IF FOO.CMD contains
%"A"
and you say
@LOAD ZOT @FOO
the exec infinitely loops typing
?Invalid CMBFP pointer
Diagnosis: At RDCMA1+4, exec does SETZM yyy(B) but unfortunately, B contains
15 because it got clobbered, so later at RDFLD0+7, SBLOCK+.CMBFP
gets clobbered.
Solution: In FILBLK, stash B away in BAKPTR as well as returning it, and
use BAKPTR at RDCMA1.
TOPS20 Change Order Number 4.2354
Written by: EKLUND 27-Jul-79 16:05:32
Edit checked: YES Document: YES
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECQU
Related TCO's:
Related SPR's: 13236
Problem: Switches appearing after commas in COMPILE-class and PRINT-class
commands do not apply to the files which follow a la TOPS-10. These file
specific switches must appear either globally (before ANY file specs) or
locally (after any particular file spec), but not in a semi-global context,
i. e., after a comma.
Diagnosis: Code tends to ignore commas, and apply any switches to the most
immediately preceding file (if any). This is clearly counterintuitive, and
misleading at best. It is most difficult, however, to force the switches
to apply to only the following file specs. Therefore, we are forced to make
all file-specific switches forbidden after a comma is seen, and allow them
only at the beginning of commands (global), or after a specific file (local).
Solution: As in the diagnosis. Observe that this also makes it illegal to
confirm a command of the above form after a comma is seen - one needs to
give a file spec after any comma seen before allowed to confirm. This
only seems right!
TOPS20 Change Order Number 4.2358
Written by: OSMAN 30-Jul-79 08:28:12
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execmt
Related TCO's:
Related SPR's:
Problem: Exec says
?Mount request FOO - error message
Diagnosis: Should be clearer
Solution: Make it say
?Mount request FOO failed - error message
TOPS20 Change Order Number 4.2359
Written by: OSMAN 31-Jul-79 14:35:53
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execmt
Related TCO's:
Related SPR's:
Problem: No colon on /PROTECTION switch in MOUNT
Diagnosis: It should have one, since it takes a value.
Solution: Use TV macro instead of T macro. (Don't confuse TV macro
with TV macro!)
TOPS20 Change Order Number 4.2360
Written by: OSMAN 1-Aug-79 09:42:51
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: Saying SET PAGE-ACCESS (OF PAGES) 123456 causes
illegal trap in exec.
Diagnosis: Exec was assuming RPACS would succeed, which it won't for ugly
page numbers.
Solution: Put ERJMP after RPACS
TOPS20 Change Order Number 4.2361
Written by: OSMAN 1-Aug-79 10:12:00
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: The command SET NO DEFAULT TAKE ECHO makes no sense
Diagnosis: Poorly designed
Solution: Don't allow SET NO DEFAULT TAKE at all.
Since TAKE always defaults to ECHO or NO ECHO, it makes no sense to allow
SET NO DEFAULT TAKE. Users can get both options via SET DEFAULT TAKE ECHO
and SET DEFAULT TAKE NO ECHO.
TOPS20 Change Order Number 4.2362
Written by: OSMAN 1-Aug-79 13:57:30
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: Defining LPT: as DSK: causes INFO STR * to list connected structure
twice.
Diagnosis: Exec loops through names in DEVNAM table via GETAB and
prints any whose DVCHR says it's a disk. LPT gets translated to a disk
and exec prints it.
Solution: Try a STDEV and see if same string comes out as went in. If not,
assume it's DSK or logical name and skip it. Hence LPT will go in and come
out as PS (or some other structure) which won't match, so LPT will get
correctly skipped! Clever, isn't it?
TOPS20 Change Order Number 4.2363
Written by: OSMAN 1-Aug-79 14:59:08
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: No INFO DEF TAKE
Diagnosis: Whoever put in SET DEF TAKE never thought of it
Solution: Put it in
TOPS20 Change Order Number 4.2368
Written by: OSMAN 2-Aug-79 09:06:33
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: TAKE subcommand mode allows NO NO NO NO...
Diagnosis: Table of what you can type after NO allows NO
Solution: Remove the NO entry from the NO table
TOPS20 Change Order Number 4.2370
Written by: DNEFF 2-Aug-79 16:48:25
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: If an illegal instruction is executed in location 777777,
the EXEC gets an error within an error. The particular case that
caused this was 777777$G typed to DDT.
Diagnosis: The routine that prints the PC of the illegal instruction
attempts to decrement the PC by one (since the hardware has
incremented it before the trap), but does so with full-word
instructions. This is contrary to the hardware, which increments
PCs MOD the size of a section. The EXEC thus ends up with an
address of 777777,,777777, or -1. This address is then passed
to LOADF and then to MAPPF, which treats a -1 as a command
to unmap the currently mapped page, and things lead down hill
from there.
Solution: Decrement the PC by 1 without a carry from the
left half of the address.
TOPS20 Change Order Number 4.2372
Written by: OSMAN 3-Aug-79 10:01:40
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execmt
Related TCO's:
Related SPR's:
Problem: Bad error message when "SKIP 1" attempted when tape already at
LEOT.
Diagnosis: There's no error handler for MTOPR failing in exec.
Solution: Put ERCAL after MTOPR and just tell user how he failed.
TOPS20 Change Order Number 4.2373
Written by: OSMAN 3-Aug-79 13:50:51
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: The command V SYSTEM:ERRMES,FOO prints wrong, assuming ERRMES is good and
FOO is bad.
Diagnosis: Exec prints error for FOO before finishing ERRMES line.
Solution: Call DFREST when error detected, to make sure previous line is
done.
TOPS20 Change Order Number 4.2374
Written by: R.ACE 3-Aug-79 17:51:10
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECMT
Related TCO's:
Related SPR's:
Problem: MOUNTED MAGTAPES CAN BREAK OLD BATCH CONTROL FILES IN RELEASE 4
Diagnosis: UNLOAD COMMAND GIVES ERROR MESSAGE (BEGINS WITH A "?") IF
DEVICE IS A MOUNTED (I.E. MT) MAGTAPE.
Solution: FIX UNLOAD COMMAND TO GIVE WARNING (I.E. "%") IN THIS CASE
TOPS20 Change Order Number 4.2378
Written by: OSMAN 8-Aug-79 09:52:35
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: Users want to know whether tape retrieval-waits are allowed
Also, some occurances of "are not" came out capitalized in INFO SYSTEM
Diagnosis: new feature needed
Solution: Add a line to INFO SYSTEM. Fix "are not" to be small letters.
TOPS20 Change Order Number 4.2380
Written by: OSMAN 10-Aug-79 08:17:50
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: Failures in RENAME command don't get noticed by batch.
Diagnosis: Error message doesn't start at left margin
Solution: Use LERROR instead of manual error printout if RNAMF jsys fails.
TOPS20 Change Order Number 4.2383
Written by: OSMAN 10-Aug-79 14:34:12
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execmt
Related TCO's:
Related SPR's:
Problem: If you say MOUNT STR FOO:, where FOO: is a logical name for structure
SNARK:, you get
Structure FOO: mounted
but we would like
Structure SNARK: mounted
which is consistent with other commands, like DIR FOO:, which will say SNARK:
in the heading.
Diagnosis: Exec uses string user typed for message
Solution: Make exec do DEVST (it already has a STDEV!) and replace string with
the real mccoy.
TOPS20 Change Order Number 4.2384
Written by: OSMAN 10-Aug-79 14:50:11
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: When doing RETRIEVE, if no files are done, there's no warning.
Diagnosis: Exec prints none
Solution: Say
%No files found for retrieving
if no files are found
TOPS20 Change Order Number 4.2385
Written by: OSMAN 10-Aug-79 15:20:57
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse EXEC2 EXECGL EXECSU
Related TCO's:
Related SPR's:
Problem: If you say SET FILE INVIS FOO followed by an escape, you don't
get to type another filespec if you want to!
Diagnosis: Code was calling INFGNS which is cool if there are guide words
that we want the escape to provoke, but for SET FILE INVIS etc. it is
inappropriate.
Solution: Create TYPFLS which is what the TYPE command uses, since SET FILE
INVIS wants the same kind of input as TYPE command. Call TYPFLS for TYPE
command, SET FILE INVIS, SET FILE PROHIBIT, SET FILE VISIBLE
TOPS20 Change Order Number 4.2389
Written by: OSMAN 13-Aug-79 09:21:55
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: No guide word after words such as BATCH in CANCEL (REQUEST TYPE) BATCH
Diagnosis: MODIFY has one. Why not CANCEL?
Solution: Use the same guide wordas MODIFY (ID).
TOPS20 Change Order Number 4.2396
Written by: DBELL 14-Aug-79 13:35:22
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem:
If take files are nested too deeply, the EXEC will not type out the
message "TAKE commands nested too deeply", but will instead say
"?Pushdown overflow internal trap".
Diagnosis:
In the routine PUSHIO, the EXEC stacks information about the current
command level. The PUSH instruction has an ERJMP after it which is
not taken.
Solution:
It would be nice to fix the monitor to make ERJMP after a PUSH which
gets an overflow work. But this is nontrivial. So for now just have
the EXEC avoid the problem by checking for overflows by hand.
TOPS20 Change Order Number 4.2399
Written by: OSMAN 15-Aug-79 11:07:11
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0 EXECSU EXECGL
Related TCO's:
Related SPR's:
Problem: When ECHOing TAKE files, you don't get the line echoed that causes
an error. Hence you don't know what failed.
Diagnosis: Echoing only occurred at command confirmation.
Solution: Do echoing at two places, one at command confirmation, and the other
at command error, invoked by CMERRX macro.
Note: In order to make the special case of
?Unrecognized command
cause the erroneous command to get echoed, it is now invoked with
CMERRX instead of ERROR. This means it will be longer, saying either
?Unrecognized command - Does not match switch or keyword
or
?Unrecognized command - File not found (or other filespec diagnostic)
TOPS20 Change Order Number 4.2400
Written by: OSMAN 15-Aug-79 14:02:58
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: In DIRECTORY command, sometimes you want EVERYTHING except certain
columns, for instance if the EVERYTHING printout will cause lines to be too
long.
Diagnosis: There's no convenient way to turn off things.
Solution: Allow the following new subcommands:
@@NO ACCOUNT
@@NO CHECKSUM
@@NO CRAM
@@NO DATES (OF) CREATION
READ
TAPE-WRITE
WRITE
@@NO DOUBLESPACE
@@NO GENERATION-RETENTION-COUNT
@@NO LENGTH
@@NO LPT
@@NO PROTECTION
@@NO REVERSE
@@NO SEPARATE
@@NO SIZE
@@NO TIMES (AND DATES OF) CREATION
READ
TAPE-WRITE
WRITE
@@NO USER
TOPS20 Change Order Number 4.2401
Written by: OSMAN 15-Aug-79 16:36:41
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: You can get
?File still mapped
when you type ^C to a new job
Diagnosis: Exec was allowing ^C too early, before it had initialized. This
bug cropped up when I reorganized exec to minimize jsyses.
Solution: Enable ^C later.
TOPS20 Change Order Number 4.2403
Written by: OSMAN 16-Aug-79 09:38:40
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: Exec is very crass about SET NO DEF COM FOO when none were set for
FOO.
Diagnosis: I was lazy
Solution: Diagnose the error, and either say
?File type or "*" expected
or
%No defaults were set for file type FOO
depending on whether FOO is null or non-null .
TOPS20 Change Order Number 4.2404
Written by: OSMAN 16-Aug-79 10:14:09
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: The SET DIRECTORY commands ask for password and needn't.
Diagnosis: none
Solution: Don't ask for password unless the monitor says one is needed
TOPS20 Change Order Number 4.2413
Written by: OSMAN 21-Aug-79 10:36:23
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: The following command fails to produce a reverse chronological
directory:
@VDIRECTORY,
@@REVERSE
@@CHRONOLOGICAL
@@
Diagnosis: Buffer the filespecs if REVF on (reverse sorting) and
don't bother to sort if alphabetic and it's a disk, since the monitor
does it for us.
Solution: Remove REVF from the definition of SORTF.
TOPS20 Change Order Number 4.2415
Written by: DBELL 22-Aug-79 15:43:33
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem:
On an ADVISE or TALK command to one of your own jobs, the EXEC
complains that there are multiple jobs to choose from. This is
incorrect, since there really is a unique job to pick, since
you cannot ever ADVISE or TALK to your own job.
Diagnosis:
Code at LINK3 in EXEC1 never weeds out our own job.
Solution:
Check for our own job, and ignore it.
TOPS20 Change Order Number 4.2426
Written by: EKLUND 28-Aug-79 13:53:46
Edit checked: YES Document: NO
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's: 13162
Problem: Minimal support for SAIL in the EXEC. In particular, LOWTSA.REL
needs to be loaded first if there is any SAIL program involved.
Diagnosis: No code to do it in the EXEC.
Solution: Add the code which keeps track of whether SAIL is being used, and
drag in LOWTSA.REL from SYS: if SAIL program seen anywhere. All the code
was given in the SPR, and is in EXECCS.MAC.
TOPS20 Change Order Number 4.2427
Written by: OSMAN 28-Aug-79 15:21:17
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1 EXECSU EXECGL
Related TCO's:
Related SPR's:
Problem: If a single DELETE command deletes files in more than one directory,
only one directory is mentioned in the
[DIR n PAGES FREED]
message.
Diagnosis: Exec is dumb
Solution: Type out the announcement for every directory instead of only one.
TOPS20 Change Order Number 4.2433
Written by: OSMAN 31-Aug-79 13:05:29
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: There's a SETM instruction that looks like a typo.
Diagnosis: It appears that the SETM IPCBFF should be a SETZM, but that
it isn't in the right place. Also, there's no SETOM IPCBFF anywhere!
Solution: Remove all references to the flag entirely. It isn't really needed,
since it signals an error condition that the exec never allows to occur, namely
that of there being no free slot into which to recieve an IPCF message. Since
the exec makes a free slot either by processing some of the messages received
already, or flushes a message (and prints a warning) the situation will never
happen.
TOPS20 Change Order Number 4.2436
Written by: OSMAN 4-Sep-79 10:41:03
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: Comments aren't allowed in indirect files used for COMPIL-class
commands.
Diagnosis: The exec parses the contents "itself" instead of through COMND
jsys, which normally does comment processing.
Solution: It would be nice to use COMND instead of old-fashioned exec code.
However, until we do...
Enhance CMPRES routine to strip carriage returns and change linefeeds to
commas. Instead of reading indirect file characters directly from file
while parsing, read the entire file into a buffer and then call CMPRES so
that the comments get correctly interpreted. (Note that CMPRES was already
being called for comments coming from the terminal).
TOPS20 Change Order Number 4.2439
Written by: OSMAN 5-Sep-79 09:32:48
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSE
Related TCO's:
Related SPR's:
Problem: QUENCH gets run on SET DIRECTORY PASSWORD command!
Diagnosis: CREDIR routine was skipping on success, which sent exec into
whatever code was next, which happened to be QUENCH code. This problem
cropped up when I removed one of two consecutive lines in EXEC0 awhile ago.
It used to say
CALL ... ;EXECUTE THE COMMAND
JRST CMDIN4 ;DONE
JRST CMDIN4 ;IN CASE SOMEBODY SKIPS
I removed the second line. Nobody should skip!!!
Solution: Print the monitor's error message if SET DIRECTORY PASSWORD fails.
TOPS20 Change Order Number 4.2440
Written by: OSMAN 5-Sep-79 09:57:15
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1 EXECP
Related TCO's:
Related SPR's:
Problem: TV can bomb out with
?JFN is not assigned
Diagnosis: Exec does its general jfn flushing after all commands including
ones that do GET jsyses. Since the monitor has already closed the jfn used
by the GET jsys, another program may now be using it (TV in this case!) so
when the exec flushes it, TV loses whatever file it was using on that jfn.
Solution: The exec should not try to close the jfn used by GETs. This is
effected by throwing the jfn off the jfn stack in the exec after the
GET jsys.
TOPS20 Change Order Number 4.2443
Written by: OSMAN 5-Sep-79 13:36:59
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: SET DIRECTORY PASSWORD fails to.
Diagnosis: CREDIR routine now takes arg in A instead of B but .DPASS fails
to set up A.
Solution: Set up A instead of B.
TOPS20 Change Order Number 4.2444
Written by: OSMAN 6-Sep-79 11:05:42
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem: TERMINAL NO FORM implies TERMINAL NO INDICATE
Diagnosis: There seems to be no reason why this is so. Perhaps someone
attempted to make TERM INDICATE mean "indicate with ^L" if the terminal
doesn't have formfeeds.
Solution: Make TERM FORM and TERM INDIC be independent.
TOPS20 Change Order Number 4.2448
Written by: OSMAN 6-Sep-79 15:09:40
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: On exec operations, file data errors don't print a very informative
message.
Diagnosis: The exec only prints the PC, not a very useful piece of information.
Solution: There are many device-dependent status situations that can
conceivably be reported. However, for now, we'll respond to SPR # 20-13102
and merely do what they asked:
1) The message will no longer include the PC.
2) The filespec will be printed that caused the error.
3) If it's a magtape and it has reached end of tape, the
exec will tell the user that.
TOPS20 Change Order Number 4.2459
Written by: OSMAN 12-Sep-79 10:57:43
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU EXECGL EXECPR
Related TCO's:
Related SPR's:
Problem: Sometimes, a magtape command given to the exec will take a real long
time because the operator has to mount another volume. The user can't ^C
out while this is happening.
Diagnosis: The exec was disabling ^C while it did its GTJFN on the tape, so
that it was assured of being able to stack the jfn before ^C happens, in order
to know to flush the jfn later.
Solution: Use CLZFFF instead of PIOFF in order to know to do CLZFF if ^C happens
in sticky spots.
TOPS20 Change Order Number 4.2466
Written by: OSMAN 13-Sep-79 16:11:53
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: When a LOAD-CLASS command indirect file specifies another indirect
file that can't be accessed, usually because it doesn't exist, or logical names
haven't been set up correctly, etc., you get
?File not found
and you don't know which one.
Diagnosis: Exec doesn't try too hard
Solution: Try to print a better error
TOPS20 Change Order Number 4.2470
Written by: TOMCZAK 14-Sep-79 15:54:53
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC2
Related TCO's:
Related SPR's: 13316
Problem: When using COPY or APPEND with long files, the message "%Unexpected
error: page table does not exist and file not open for write
proceeding..." sometimes appears.
Diagnosis: 1) a file with a section long hole in it and
2) a file whose EOF pointer points to the last page of a section
during some copy operations (eg disk-nondisk) and APPENDS, EXEC attempts
to optimize it's time by prefaulting pages from the file. if the set of pages
it wants to preload are in a non existent section, the preload operation
fails but causing no problems for the continuation of the COPY/APPEND.
Solution: before preloading pages--
1)check for non-zero section
2)see if pages have already been preloaded from this section
3)check to see that the section needed really exists
NOTE: long files with section long holes in them will be doing an extra FFUFP
JSYS for EVERY PAGE in that section.
TOPS20 Change Order Number 4.2471
Written by: TOMCZAK 17-Sep-79 14:26:37
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's: 13334
Problem: Lower forks can get a JFN that an EXEC above it is using for a TAKE
command and mess around with (EG close it). Upon returning to the upper level
EXEC, if the JFN for the command file or the log file is cloased, EXEC throws
up. This is particularly bad when the EXEC itself is put in a lower fork (as
in a PUSH command within a TAKE file). The CLOSE command in particular will
simply close the JFNs the upper EXEC is using.
Diagnosis: A lower EXEC, in executing a CLOSE command, simply loops around
looking for valid JFNs. It finds the JFNs for the upper EXEC's TAKE command
and closes them, unaware of the special status of these JFNs.
Solution: Whenever JFNs are obtained for the TAKE command, make them restricted
to the current fork or higher by turning on the GJ%ACC in the COMND call.
TOPS20 Change Order Number 4.2472
Written by: OSMAN 17-Sep-79 16:32:24
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: The command
@TAKE FOO NUL:
causes "JFN is not assigned" error
Diagnosis: NOTIO is broken.
Solution: Fix NOTIO. (It seems to have always been broken!)
TOPS20 Change Order Number 4.2474
Written by: TOMCZAK 18-Sep-79 12:28:39
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's: 13370
Problem: /logname: switch to SUBMIT command hadles the log file differently
from the way default log file is handled. Currently will always overwrite it.
Diagnosis: If the /logname: switch is not used, EXEC passes BATCON a filespec
built from the connected structure/directory the CTL filename and .LOG for the
extension. No generation number is passed. When the /logname: switch is used,
the filespec parsed will be passed to BATCON, including the generation number.
Since EXEC is parsing a filespec with the GJ%FOU bit on, BATCON will always
be getting a filespec on a new file instead of the original.
Solution: 1) Don't pass the generation number to BATCON.
2) Have EXEC parse the log filespec so that it will recognize an existing
file or allow creation of a new one. (No GJ%FOU bit in GTJFN call)
TOPS20 Change Order Number 4.2475
Written by: TOMCZAK 19-Sep-79 14:13:56
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's: 13274
Problem: HELP * DOESN'T HANDLE THE KEYWORD LISTING AS INTELLIGENTLY AS HELP ?
Diagnosis: EXEC IS TRYING TO SIMULATE WHAT THE COMND JSYS DOES ON A QUESTION
MARK BUT DOESN'T CHECK THE LENGTH OF EACH KEYWORD IN THE KEYWORD TABLE BEFORE
TRYING TO PRINT THEM. THIS CAUSES THE COLUMNS TO NOT LINE UP CORRECTLY.
Solution: GET RID OF THE PARSING OF THE STAR (*) SINCE IT DOES ALMOST THE SAME
THING AS THE QUESTION MARK (?).
TOPS20 Change Order Number 4.2479
Written by: OSMAN 20-Sep-79 14:14:52
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: The FIND subcommand feature of DIRECTORY-class commands is broken.
Diagnosis: The particular algorithm the exec uses to do FIND depends on
the filespecs being buffered. A recent "improvement" in the DIRECTORY
command involves printing specs immediately instead of buffering. This
breaks FIND.
Solution: For now, merely add FIND as the list of options requiring buffering,
along with CHRONOLOGICAL. A future improvement might be to change the FIND
algorithm so as not to require buffering.
TOPS20 Change Order Number 4.2481
Written by: OSMAN 21-Sep-79 12:55:18
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC3
Related TCO's:
Related SPR's:
Problem: The TDIRECTORY command gets screwed up if the directory is so large
that the message
STORAGE FULL,
DIRECTORY WILL BE PRINTED IN TWO OR MORE SECTIONS
results.
Diagnosis: There is a longstanding bug that causes the sort routine to
sort one more word than is in the table. This bug didn't used to show up
if we ran out of fdb buffer space before the sort table got used up. However,
now that the sort table exhausts first, we see the bug.
Solution: Check for length sorted table, instead of looking for a 0 beyond
the end of the table.
TOPS20 Change Order Number 4.2498
Written by: OSMAN 28-Sep-79 08:35:43
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: Exec doesn't print double colons under INFO DEFAULT PRINT etc.
Diagnosis:
Solution: Put in the double colons in the display
TOPS20 Change Order Number 4.2499
Written by: OSMAN 28-Sep-79 08:37:29
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execqu
Related TCO's:
Related SPR's:
Problem: spurious 4 at beginning of edit line in EXECQU!
Diagnosis: It doesn't hurt, but uses an extra word of storage.
Solution: remove it
TOPS20 Change Order Number 4.2500
Written by: OSMAN 28-Sep-79 08:47:02
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0
Related TCO's:
Related SPR's:
Problem:
@DEFINE DBELL: DSK:
@SYSTAT DBELL:
fails!
Diagnosis: DBELL matches user name, and then colon causes confusion.
Solution: COMND has a general problem that if the first few characters of the
input matches some field in the chain fed by the program, and gobbling more
characters matches a field further down the chain, the first field matched
is opted for.
Fix COMND to see which field in the chain matches most input and use that
one. So DBELL: is a better match for directory than user name, since the colon
can be included for directory name. Hence make COMND match DBELL: with .CMDIR,
regardless of where .CMDIR is in the chain with respect to .CMUSR.
The old way was bad because in order to get recognition of user names, you
have to put .CMUSR before .CMDIR, since "DBE<esc>" will be ambiguous as a
directory name, but recognizable as "DBELL" as a user name. On the other hand
we'd like to put .CMDIR first so that "DBELL:" gets interpreted as a directory
name.
The new way gets it all right!
TOPS20 Change Order Number 4.2509
Written by: OSMAN 3-Oct-79 19:57:55
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECCS
Related TCO's:
Related SPR's:
Problem: If you start up a fresh exec and say
@LOAD ?
the help message correctly lacks carriage return as an option,
since no default string has been set up yet.
However, if you say
@LOAD /SEARCH !any switch works in this example! ?
then carriage return is advertised as a possibility.
Diagnosis: The check for whether to allow carriage return was not made at
a low-level enough spot.
Solution: Make the check in COMIN. Note that the entire TI routine doesn't
always get it right. It assume that indirect files have filespecs
in them, and hence cr is allowed after an indirect spec. If the
assumption is wrong, the infamous
?No saved arguments
will be your prize.
TOPS20 Change Order Number 4.2510
Written by: OSMAN 4-Oct-79 10:41:32
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: On EOF interrupt, wrong PC is printed
Diagnosis: Exec was clobbering LEVTAB
Solution: Fix EOFPSI routine
TOPS20 Change Order Number 4.2518
Written by: OSMAN 8-Oct-79 15:30:52
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC0 EXEC1 EXECGL
Related TCO's:
Related SPR's:
Problem: LIST commmand exists
Diagnosis: There didn't used to be PRINT
Solution: Remove LIST. If anyone wants any of its whizzy features in the
future, put them in the galaxy PRINT system.
TOPS20 Change Order Number 4.2519
Written by: OSMAN 8-Oct-79 15:50:18
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's:
Related SPR's:
Problem: ^T seems not to work when redirecting TAKE output to a file.
Diagnosis: The ^T output is going into the file.
Solution: Send ^T output to .PRIOU. This is more consistent with the fact
that the ^T was typed from the terminal.
TOPS20 Change Order Number 4.2520
Written by: OSMAN 8-Oct-79 16:12:19
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's:
Related SPR's:
Problem: ^H after CONNECT command with password displays the password.
Diagnosis: Exec was forgetting to wait for confirmation after password. Hence
COMND thought there was an error.
Solution: Remind the exec to wait for confirmation as PASLIN.
TOPS20 Change Order Number 4.2521
Written by: OSMAN 9-Oct-79 11:59:03
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: The user name lists under INFO STRUCTURE look messy
Diagnosis: The exec was doing HRROI A,DRRBUF and then using A as a byte pointer
with which to measure the length of the next user name to see if it would
fit on the line. Unfortunately, ILDB doesn't understand -1,,ADR.
Solution: Change HRROI to MOVEI followed by HRLI of 440700.
TOPS20 Change Order Number 4.2541
Written by: OSMAN 22-Oct-79 13:44:06
Edit checked: NO Document: YES
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execin
Related TCO's:
Related SPR's:
Problem: If JFN gets released while INFO FILS is looking, exec's jsys can
fail, causing strange error message.
Diagnosis: A CONTINUE STAYed program can cause such a situation!
Solution: Put ERJMPs at appropriate places and analyze the error. If a jfn
gets released while INFO FILS is in progress, and the exec detects this
because a jsys fails, the line of output will be abridged wherever the jsys
fails, and the rejoinder
... [JFN has just been released]
will end the line.
TOPS20 Change Order Number 4.2544
Written by: TOMCZAK 24-Oct-79 15:31:14
Edit checked: NO Document: NO
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXEC1
Related TCO's: 4.2545
Related SPR's:
Problem: If SET DEFAULT TAKE ECHO is set, then
@TAKE FOO,
@@NO ECHO
@@
has no effect
Diagnosis: The handling of flags PECHOF and ECHOF is inconsistent
Solution: Set an AC at the start of .TAKE in EXEC1 to the value of PECHOF
and then reset that AC during execution of the subcommand so that ECHOF
can be set to the proper value just before exiting from the TAKE code.
TOPS20 Change Order Number 4.2545
Written by: TOMCZAK 24-Oct-79 15:33:57
Edit checked: NO Document: NO
TCO Tested: YES Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECSU
Related TCO's: 4.2544
Related SPR's:
Problem: Echoing of commands in a TAKE file don't work
Diagnosis: Handling of PECHOF and ECHOF flags is inconsistent
Solution: In ECHCMD test only the ECHOF flag and make sure it is
properly set in the TAKE command itself. TCO# 4.2544
TOPS20 Change Order Number 4.2554
Written by: TOMCZAK 26-Oct-79 11:20:43
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: EXECIN
Related TCO's:
Related SPR's:
Problem: INFORMATION (ABOUT) TERMINAL DOESN'T WORK RIGHT WHEN AN ILLEGAL
TERMINAL TYPE HAS BEEN SET
Diagnosis: WRONG AC BEING USED IN XCT INSTRUCTION AT ILTTYP+1
Solution: CHANGE (B) TO (A) IN XCT INSTRUCTION
TOPS20 Change Order Number 4.2589
Written by: OSMAN 5-Dec-79 10:23:13
Edit checked: NO Document: NO
TCO Tested: NO Maintenance Release: NO
Hardware-related: NO
Program: EXEC
Routines Affected: execse
Related TCO's:
Related SPR's:
Problem: EXECSE doesn't assemble without an error. You get
DEFINE UNASSIGNED, DEFINED AS IF EXTERNAL
Diagnosis: Symbol $DEFAU conflicts with something, probably in GLXLIB.
Solution: Change $DEFAU to TDEFAU