Miscellaneous HP3000 Programming Bugs and Features
A :FILE LOADLIST=$NULL can re-route all loader error messages to $NULL, since they seem
to ignore HPMSGFENCE settings and other controls.
How can you get the #%@$# cursor keys to work? With TERM='hp2392a' they don't. vi is
just too weird without 'em :-)
LB
Posix vi looks for an EXRC environment variable, which by default is "./.exrc".
Put mappings like the ones shown below in that file to map your arrow keys to the proper
vi key. Note -- use a real escape character where you see "^[".
map ^[A k
map ^[B j
map ^[C l
map ^[D h
CL
I'm assuming you're on an XL system. Just call the COMMAND Intrinsic and pass it the MPE
command PAUSE.
RAT
Yes!
From my stint at the HP Response Center (greetings to my former fellows!), I've got
several sample COBOL routines to do this. Here's my favorite one, which uses HPEXTIN to
convert a string to a REAL value:
(much COBOLeez omitted)
WORKING-STORAGE SECTION.
01 string-num pic x(5) value "00005".
01 string-len pic s9(4) comp value 5.
01 decplaces pic s9(4) comp value 0.
01 datatype pic s9(4) comp value 1.
01 scale pic s9(4) comp value 0.
01 delims pic s9(4) comp value 0.
01 pause-value pic x(4) value "xxxx".
01 error1 pic s9(4) comp value 0.
01 d-error pic 9(5) value 0.
(somewhere in PROCEDURE DIVISION land )
CALL INTRINSIC "HPEXTIN" USING string-num, string-len,
decplaces, datatype, scale, delims, pause-value, error1.
if cc not equal zero
display "Error on HPEXTIN call"
move error1 to d-error
display "Error = " d-error
goback.
CALL INTRINSIC "PAUSE" USING pause-value.
To change the value, simply change the text in "string-num" to the number of
seconds.
There are several examples of this on HP SupportLine as well.
TBS
Try this (from my infamous 3-ring binder:
{ Cute trick for PAUSE intrinsic:
{ 01 xx.
{ 03 high-word s9(4) comp value %040100.
{ 03 low-word s9(4) comp zero.
{ * Evaluates to x'40400000' or 1 second, +1077936128.
{ * For 5 seconds, +1083179008, x'40900000'.
RAT
I've had good luck with:
:FILE term,NEW;DEV=nnn;ACC=INOUT
This allows both read and write to the ldev (if it's AVAILable) and doesn't generate a
REPLY request on the console.
Also be sure to check the Condition Code after the FOPEN to ensure it actually worked.
Call PRINTFILEINFO if it's not CCE and that should give you a good FSERROR number to
lookup.
JV
The variable HPCWD exists on 4.5 and later
The command NEWLINK exists on 5.0 and later (there must be a better way...)
listfile ;seleq=[object=hfsdir] 'hfsdir' may not be supported on 4.5, definitely isn't
on 4.0.
NOTE:
On the Express 3 release of 5.0 (June?) there are 2 new CI variables: HPOSVERSION and
HPRELEASE version. These contain the os and release ids seen in the :SHOWME banner. Eg.
:showvar hp@version
HPOSVERSION = B.79.06 (can't be changed)
HPRELVERSION = B.79.05 (can't be changed)
HPVERSION = B.79.05 <== this can be changed in sysgen.
GC
When the MPE/XL machines came out, HP changed the :PURGE command to first "wipe
out" the file, effectively "shredding" the contents so that even low-level
tools could not recreate its contents. (I believe this was in response to a DoD
requirement.) For large files, this can take a long time. However, by first removing write
access to the file, the :PURGE command cannot write to the file, but the file is deleted
anyway! The following command file shows how this works.
PARM filename="?"
if "!filename" = "?"
echo Function: Purge a large file quickly
echo Syntax : PURGEQ filename
echo Parm : filename - the name of the file to be purged
echo Example : PURGEQ BIGFILE
echo
return
endif
FILE X = !filename, OLD; ACC=IN
if finfo( "*X", "exists" )
PURGE *X
else
PURGE !filename
endif
RESET X
The ",OLD" designation is used on the :FILE equation so that
"finfo" will not acknowledge a temporary file of the same name. The
"ACC=IN" is what removes write access.
If the file does not exist, we want :PURGE to return the appropriate error message
without reference to file "X".
No doubt other enhancements are possible, but this gets the idea across.
Incidentally, temporary files are not affected by the "long purge."
PHC
1. Use listf -3 on the file to find out the actual value of the 'PRIV'
file code.
:LISTF I0038,-3
********************
FILE: I0038.DATA....
FILE CODE : -400 FOPTIONS: BINARY,FIXED....
BLK FACTOR: 1 CREATOR : MGR
...
VOLCLASS : MPEXL_SYSTEM_VOLUME_SET:DISC
2. Issue a file equation such as:
FILE A=I0038;DEL
3. Using the following DSCOPY syntax (an error will occur but the file
will be deleted):
DSCOPY *A;$NULL;FCODE=-400
:DSCOPY *A;$NULL;FCODE=-400
Source file: I0038.DATA.....
Target file: $NULL
UNABLE TO ACCESS TARGET FILE/DEVICE. (NS/NFTERR 48)
END OF SUBSYSTEM
:LF I0@,2
LISTF I0@,2
NO FILES FOUND IN FILE-SET (CIWARN 431)
GAH
Another suggestion: The PURGEGROUP command will purge a group (as long as you have at
least AM) even if it includes PRIV files. (of course, if you have other files that you
*DON'T* want purged in that group, you'll have to copy them to a group temporarily).
GSA
Gilles Schipper (Gilles_Schipper@MAIL.MAGIC.CA) adds:
:FILE T=$NULL
:STORE privfilename;*t;SHOW;PURGE
This works on 5.0. For 4.0 or earlier, I believe you cannot use a NULL file equation
for the tape. It should work if you use a real (or is it reel) tape.
In either case, you will need the appropriate capability to store PRIV files (i.e. OP
or PM).
Alternatively, you can run FSCHECK.MPEXL.TELESUP and use the PURGEFILE command.
GS
for 4.0, simply add ;TRANSPORT to the :STORE command. CMSTORE has always supported storing
to $NULL. It's just NMSTORE/TURBOSTORE that didn't support it until 5.0.
Of course if it is a big file, you probably have to wait a while as it writes all the
data out to $NULL before purging your file.
JK
Want to know if you're a VT session initiated by a batch job's REMOTE HELLO? Well,
HPINTERACTIVE is true (even if started by batch job), so try to set HPTYPEAHEAD true. If
it fails, your origin is a batch job.
JK
Do a NSINFO of the $BACK environment.
If you get an error, you're not a VT session. If no error, you get back the name of the
originating node (usually :-) ).
| 
HP3000/HP
3000 FAQ