I had a workstation today that would not install the 6.0_update32 JRE, getting the error “error 2753 regutils.dll”.

Sun/Oracle’s ‘troubleshooting’ is worthless.  Nobody else’s was any help, either, though.  JavaRa gave it a good try. (Seriously, you just have a comprehensive list of registry keys and files/directories to delete.  That’s all.)

So, fire up procmon, include “msiexec.exe” and see what pops up….

Simple enough.  The installer thinks there’s another conflicting existing installation:

Delete the registry key (and subkeys):

HKEY_CLASSES_ROOT\Installer\Products\4EA42A62D9304AC4784BF238120632FF

Grand.

So, why can’t Sun/Oracle release a utility to clean up all traces of Java?  Laziness, pure and simple.  This has been a pain for admins for years.  Fix your installer.

cf the .NET Framework Cleanup Tool

The SAP ATP (Available To Promise) server caches availability checks in the export/import buffer, replacing database calls to RESB and VBBE.

Reference:

• SAP Note 99999 – ATP server: Installation and sizing (https://service.sap.com/sap/support/notes/99999)
  
• SAP Note 702728 – Profile parameter for export/import buffer instances (https://service.sap.com/sap/support/notes/702728)
  
• SAP Note 373986 – Overflow of the export/import buffer (https://service.sap.com/sap/support/notes/373986)
  
• “SAP Performance Optimization Guide (6^th Edition),” Thomas Schneider, Galileo Press (2009), section 10.4, pp. 443-450. (http://www.sap-press.com/products/SAP-Performance-Optimization-Guide-%286th-Edition%29.html)
  

Install the ATP Server

Pick a suitable SAP Instance to run the ATP server, subject to the space constraints below.

The ATP server is a logical service activated by a profile parameter, and since all of the instance’s servers will use this same server, the parameter should be set in the DEFAULT profile (transaction RZ10). The ATP server and Enqueue server should run in the same instance.

Set both:

• rdisp/enqname
  
• rdisp/atp_server
  

Restart the instances to pick up the changes.

Tune the Export/Import Buffer

Figure out how many objects you plan to cache. Either follow the elaborate formulas in the references above, or just flush the cache, run a sample test, and extrapolate. Quick and dirty guideline: figure out how many maximum objects you’ll ever need (rsdb/obj/max_objects) and set it, and set the Export/Import Buffer (size in KB) to twice that number (rsdb/obj/buffersize), since each object should take a little less than half a KB. Tune as needed.

In ST02′s “Detailed Analysis” menu, select “Buffered Objects” to see individual records (Table Name = “ATPSB”). See transaction ACDB, as well. You can view or delete individual, or all, entries.

Transactions

• RZ10 – Set parameters.
  
• ST02 – View buffer statistics.
  
• SM50/SM51 – Verify ATP and Enqueue servers activated.
  
• ACDB – View and delete export/import buffer table entries.
  

  
  

… because I know what the girl on the right is thinking …

For posterity, here’s an edited picture that doesn’t make the girl on the left have pointy ears.

There went my lunch hour.

% minizinc product mix model from http://www.solver.com/stepbystep.htm
% 2012 Jason Filley (jason@snakelegs.org)

%%%% parameters
int: glue = 5800;		% glue (quarts)
int: pressing = 730;	% pressing time (hours)
int: pine = 29200;		% pine chips (pounds)
int: oak = 32500;		% oak chips (pounds)

constraint assert(glue >= 0,"Invalid datafile: " ++
	"glue must be non-negative.");
constraint assert(pressing >= 0,"Invalid datafile: " ++
	"pressing must be non-negative.");
constraint assert(pine >= 0,"Invalid datafile: " ++
	"pine must be non-negative.");
constraint assert(oak >= 0,"Invalid datafile: " ++
	"oak must be non-negative.");

%%%% decision variables
var 0..100: tahoe;
var 0..100: pacific;
var 0..100: savannah;
var 0..100: aspen;

%%%% constraints

% glue
constraint  50*tahoe + 50*pacific + 100*savannah + 50*aspen <= glue;

% pressing time
constraint  5*tahoe + 15*pacific + 10*savannah + 5*aspen <= pressing;

% pine chips
constraint  500*tahoe + 400*pacific + 250*savannah + 200*aspen <= pine;

% oak chips
constraint  500*tahoe + 750*pacific + 250*savannah + 500*aspen <= oak;

%%%% objective function -- maximize profit
var 0..100000: profit;
profit = 450*tahoe + 1150*pacific + 800*savannah + 400*aspen;
solve maximize profit;

output ["Tahoe: ", show(tahoe), "\n",
		"Pacific: ", show(pacific), "\n",
		"Savannah: ", show(savannah), "\n",
		"Aspen: ", show(aspen), "\n",
		"Profit: ", show(profit), "\n"];

% minizinc.exe product_mix_wood_paneling.mzn
% Tahoe: 23
% Pacific: 15
% Savannah: 39
% Aspen: 0
% Profit: 58800
% ----------

I routinely browse the openbsd-cvs mailing list, and I saw this easy openbsd-cvs bug fix (“Fix a stupid bug in tcpdump print-bgp.c“) the other night when doing some really late-night, partial-involvement sysadmin work.  So I decided to pass the time (“stay awake”) by doing a regex exercise to find similar patterns in the OpenBSD source tree.

In C:

• “a = b” (single equals sign) is the assignment operator.  Take the value of “b” and put it in “a.”
• “a == b” (double equals signs) is an equality test.  If “a” equals “b” then the expression evaluates true.

It’s an easy bug to introduce, using a single = when you want two, so I tried to find more.  Mainly, it was regex practice.

#!/usr/bin/perl -w

use strict;

if ( $#ARGV < 0 ) {
    die "Specify a file.\n";
}

my $file = $ARGV[0];

open( SOURCEFILE, "<$file" ) || die "Can't open file $file.\n";

while (<SOURCEFILE>) {

    my $line = $_;
    chomp($line);

    if ( $line =~ /^\s*if\s.*\s=\s(-)?\d/ ) {

        $line =~ s/^\s*//;s/\s*$//;

        unless ( $line =~ /^(\*|\/\*)/ ) {
            unless ( $line =~ /(\(.*\).*\s=\s)/ ) {
                print "/*$file: $. */\n$line\n\n";
            }
        }
    }
}

close(SOURCEFILE);

It includes lines that are part of a multi-line comment, but my time and interest are exhausted, and I could have sworn there were compiler options to test for this anyway …? And I don’t do C.

In the whole tree, I ended up with only a few cases where an assignment occurs in the condition of an if statement:

$ find /usr/src -type f -name \*.c -exec ./assigntest.pl '{}' \;

/*/usr/src/sys/arch/alpha/alpha/in_cksum.c: 101 */
if ((offset = 3 & (long) lw) != 0) {

/*/usr/src/sys/dev/gpio/gpiodcf.c: 466 */
if ((ymdhm.dt_year = 2000 + FROMBCD(year_bits)) > 2037) {

/*/usr/src/sys/dev/usb/udcf.c: 669 */
if ((ymdhm.dt_year = 2000 + FROMBCD(year_bits)) > 2037) {

/*/usr/src/sys/net/zlib.c: 3830 */
if ((f = 1 << (j = k - w)) > a + 1)     /* try a k-w bit table */

/*/usr/src/bin/ed/main.c: 603 */
if ((garrulous = 1 - garrulous) && *errmsg)

/*/usr/src/bin/rmail/rmail.c: 188 */
if ((from_path = malloc(fptlen = 256)) == NULL)

/*/usr/src/gnu/egcs/libio/floatconv.c: 1878 */
if (j = 11 - hi0bits(word0(d2) & Frac_mask))

/*/usr/src/gnu/usr.bin/binutils/gas/config/tc-i860.c: 606 */
if ((c = 10 * (c - '0') + (*s++ - '0')) >= 32)

/*/usr/src/gnu/usr.bin/binutils/gas/config/tc-sparc.c: 2052 */
if ((c = 10 * (c - '0') + (*s++ - '0')) >= 32)

/*/usr/src/gnu/usr.bin/binutils/gas/config/tc-tic4x.c: 464 */
if (flonum.low > flonum.leader  /* = 0.0e0 */

/*/usr/src/gnu/usr.bin/binutils/gdb/ns32knbsd-nat.c: 297 */
if (enter_addr = 0)

/*/usr/src/gnu/usr.bin/cvs/zlib/inftrees.c: 218 */
if ((f = 1 << (j = k - w)) > a + 1)     /* try a k-w bit table */

/*/usr/src/gnu/usr.bin/gas/config/tc-i860.c: 499 */
if ((c = 10 * (c - '0') + (*s++ - '0')) >= 32) {

/*/usr/src/gnu/usr.bin/gas/config/tc-sparc.c: 745 */
if ((c = 10 * (c - '0') + (*s++ - '0')) >= 32) {

/*/usr/src/gnu/usr.bin/gcc/gcc/testsuite/gcc.c-torture/execute/20001221-1.c: 4 */
if (! (a = 0xfedcba9876543210ULL))

/*/usr/src/gnu/usr.bin/gcc/gcc/testsuite/gcc.dg/20020426-2.c: 103 */
if ((f = 1 << (j = k - w)) > a + 1)

/*/usr/src/gnu/usr.bin/gcc/gcc/testsuite/gcc.dg/c99-bool-1.c: 207 */
if ((u = 2) != 1)

/*/usr/src/gnu/usr.bin/gcc/gcc/testsuite/gcc.dg/compare4.c: 36 */
if (u < (x = -1)) /* { dg-warning "signed and unsigned" "MODIFY_EXPR" } */

/*/usr/src/gnu/usr.bin/gcc/gcc/testsuite/gcc.dg/compare4.c: 38 */
if (u < (x = 10))

/*/usr/src/gnu/usr.bin/gcc/gcc/testsuite/gcc.dg/compare4.c: 40 */
if ((x = 10) < u)

/*/usr/src/gnu/usr.bin/binutils-2.17/gas/config/tc-i860.c: 606 */
if ((c = 10 * (c - '0') + (*s++ - '0')) >= 32)

/*/usr/src/gnu/usr.bin/binutils-2.17/gas/config/tc-sparc.c: 2102 */
if ((c = 10 * (c - '0') + (*s++ - '0')) >= 32)

/*/usr/src/gnu/usr.bin/binutils-2.17/gas/config/tc-tic4x.c: 463 */
if (flonum.low > flonum.leader  /* = 0.0e0 */

/*/usr/src/gnu/usr.sbin/sendmail/rmail/rmail.c: 228 */
if ((from_path = malloc(fptlen = 256)) == NULL)

/*/usr/src/lib/libc/gdtoa/dtoa.c: 212 */
if (( j = 11 - hi0bits(word0(&d2) & Frac_mask) )!=0)

/*/usr/src/lib/libc/gdtoa/gdtoa.c: 211 */
if ( (j = 11 - hi0bits(word0(&d) & Frac_mask)) !=0)

/*/usr/src/lib/libc/gdtoa/gdtoa.c: 344 */
if ( (j = 11 - hi0bits(word0(&d) & Frac_mask)) !=0)

/*/usr/src/lib/libc/gdtoa/strtod.c: 74 */
if (!scale || (i = 2*P + 1 - ((word0(x) & Exp_mask) >> Exp_shift)) <= 0)

/*/usr/src/lib/libc/gdtoa/strtod.c: 532 */
if (scale && (j = 2*P + 1 - ((word0(&rv) & Exp_mask)

/*/usr/src/lib/libedit/common.c: 909 */
if (tmplen < 0 || (tmpbuf[tmplen] = 0, parse_line(el, tmpbuf)) == -1)

/*/usr/src/usr.bin/apply/apply.c: 146 */
if ((c = malloc(clen = 1024)) == NULL)

/*/usr/src/usr.bin/xlint/lint1/decl.c: 1913 */
if (!redec && !isredec(dsym, (warn = 0, &warn))) {

/*/usr/src/usr.bin/xlint/lint1/func.c: 306 */
if (!isredec(fsym, (warn = 0, &warn))) {

/*/usr/src/usr.bin/xlint/lint1/tree.c: 3284 */
if (!eqtype(tp, tn->tn_type, 1, 0, (warn = 0, &warn)) || warn)

They look fine to me, but all the same, it’s still a curious idiom to occur so few times.

#toy test
#include <stdio.h>

int main(){
    int     a;
    if ((a = 2000 + 50) > 2037) {
        printf("TRUE - a is %d\n",a);
    }
    else {
        printf("FALSE - a is %d\n",a);
    }
return 0;
}
/* TRUE - a is 2050 */

And then I went to bed.

Note that “AtpTable” is listed as input (non-optional). Just pass it an empty string. For the output, you have to handle 3 cases: empty, 1 result (returns a single object), more than 1 result (returns an array). Make sure SOAP_SINGLE_ELEMENT_ARRAYS is set, which returns an array of 1 item when only 1 item exists (PHP considers that a feature, not a bug). Then you just test that the array exists (isset).

<?php
$SOAP_OPTS = array( 'login' => '<THEACCOUNT>',
                    'password' => '<THEPASSWORD>',
                    'features' => SOAP_SINGLE_ELEMENT_ARRAYS);
$WSDL = "http://sap.corp.example.com:8000/sap/bc/srt/wsdl/bndg_BLAHBLAHBLAH/wsdl11/allinone/standard/document?sap-client=100";
$client = new SoapClient($WSDL,$SOAP_OPTS);
$params = array(
    'AtpTable' => "",
    'Branch' => "1",
    'CheckRule' => "A",
    'GetRows' => "100",
    'IncludeUnsavSo' => "X",
    'StartMaterial' => "12345",
    'Sloc' => "0001"
);

try
{
    $result = $client->ZfmGetAtpTest($params);
}

catch (SoapFault $exception)
{
    print "***Caught Exception***\n";
    print_r($exception);
    print "***END Exception***\n";
    die();
}

#print_r($result);
if (isset($result->AtpTable->item)) {
    foreach ($result->AtpTable->item as $row) {
        print "$row->Material\t$row->Branch\t$row->AtpQty\n";
    }
}
?>

Lillian was an old woman.  She had liver spots the size of quarters.  She had Coke-bottle glasses [Literally.  They were thick.].  She was diabetic and had one leg amputated above her knee, and she was, of course, in a wheelchair.  She was a very sweet woman, but her faculties had left her, for the most part, and she wasn’t too keen about what was going on.

Her husband John was an old, overall-wearing farmer, and he came in every day before lunch.  He’d wait while the nurses assistants got Lillian ready for lunch, sitting out by Lillian’s table.  He’d help her eat lunch, and then he’d spend the day with her.  She’d nap, or watch the kids come in for piano recitals, or play Bingo, or get her fingernails done, or roll her wheelchair around, or nap, or just watch Lawrence Welk, or just spend the afternoon doing nothing.

John helped Lillian eat supper, and when the nurses assistants came to take her to bed, John took his van and went home.

Every day.

One day, John was sitting out at the lunch table, waiting for the assistants to bring Lillian out, and one of the other women, Lenora, asked John, “John, you know she would never know if you didn’t come.  You have so many other things you could be doing.  Why on earth do you come here every day?”

And John said, “Because I said, ‘For better, or for worse.’  And I meant it.”

Well, Kristi, I said for better or for worse, and I meant it.  Happy 15th anniversary.  We’ll see who’s pushing who around the nursing home, later…..

Well, two days later, Drudge linked to “Japan invents speech-jamming gun that silences people mid-sentence“.

“The device works by recording its target’s speech then firing their words back at them with a 0.2-second delay, which affects the brain’s cognitive processes and causes speakers to stutter before silencing them completely.”

So I’m not some weirdo who gets distracted by echoes.  It’s science. I feel better.  (Unless it doesn’t actually work, of course.)

Now, wouldn’t it be funny to have some iPhone/Android/whatever malware that did the same thing, rendering people unable to speak into their own phones?  They could speak as long as they’re not listening, too, turning the phone into a walkie-talkie.

That’s as funny as my relative when 3-way calling first came out.  He’d call a Chinese restaurant and a Vietnamese restaurant at the same time and let them fight over who called who.  “You call me! I no call you!”