Posts by JQW 49 publicly visible posts • joined 21 Mar 2012
I came across a review of a similar system a few months back that came with a Volume License Key installation of Windows 10. They had also hacked the Windows installation to prevent Windows Update from automatically running to stop it phoning home and getting the installation of Windows invalidated.
1994 or so, I was tasked with setting up our office's internet connectivity, and needed to set up an internal DNS server. The first choice was to use a spare SCO Unix system, but the DNS server implementation running on that was horribly out-of-date. The second option was to install a DNS process on a server running the core OS we sold, but the only DNS implementation for that just wasn't yet up to scratch.
So I grabbed a surplus 386SX machine which had previously been running Windows 3.1, downloaded some Slackware Linux floppy images, and had a working DNS server up and running to tide us over until I could source something better.
It was still running when the company went bust around 2000, surviving an office move and merger in the process.
At a former workplace we acquired a job lot of discontinued old servers as there was still demand for spares. These were Intel processor based, but not PC architecture, and only ran one obscure and forgotten operating system. I forget the precise details, but there were two models of these servers, with memory expansion boards for one model retailing at about four times the price of the other.
We stripped these servers down to extract the memory boards, and noticed that they were identical, right down to the model number. Further examination of the recovered boards revealed an un-marked option jumper, set one way on the cheaper boards, and in the opposite way on the expensive ones.
The same vendor charged had also charged a 400% premium on SCSI drives - the hardware would only recognise drives that had been pre-formatted with a special signature. We found a fix for that - a hidden menu option on the system's diagnostic tape.
A former workplace used to own a high-end PC workstation, manufactured around 1988 or so, but used occasionally a few years later. It was a fairly impressive piece of equipment, relatively small in size, but powerful enough for CAD or other uses, yet small enough for a monitor to sit directly on top of the box. The manufacturer was a long forgotten UK mini-computer company who were attempting to branch out into the desktop market.
There was just one major problem. The machine's power on switch was a red square push button situated in the middle of the front bezel. At keyboard height. And not recessed. Oh dear!
I have managed to get it to build. The documentation isn't that great, and the build process has a choice of several different PDP-10 simulators to build as part of the process, each with different pre-requisites. Even once every required component is installed, the build process could time-out, but I gather these problems have been somewhat alleviated.
The last time I tried to build it I had other probems - although the ITS installation itself built fine, I had issues building one of the extra graphical terminal emulators which I couldn't resolve at the time.
I used to work with a clunky E-mail system which would only show the 200 most recent E-mails in any folder. Everything else was effectively invisible until you deleted or moved particular messages. This affected both the standard bundled DOS application and the several third-party Windows applications that appeared. Perhaps this chap had used that one in the past?
This reminds me of a support call I took a couple of decades ago. One of the products we sold was a long forgotten E-mail client that was an optional extra to the equally forgotten E-mail system.
The vendor had released a new Windows version which was very, very buggy, with a tendency to crash with 'Assertion Failed' errors at an inopportune moment when processing any rule that worked under the previous version. It was obvious that this software had shipped months, if not years, before it was ready.
I took a call from one client who had a site running the French translation of this client, which was failing to empty the wastebasket folder. Odd, because the underlying mail API just moved waste mail into a dedicated folder on the server, and a configurable job would run on the server at a set time (usually 3 a.m.) and empty every wastebasket at once. Obviously the new version of this mail client now did wastebasket handling too, and somehow that was also broken.
After a bit of fiddling, we discovered that this French translated client had two wastebasket folders, named 'Corbeille' and 'Poubelle'. Mail got moved into one folder when deleted, but the automatic rule for emptying the wastebasket operated on the other one! If errors like that happen, no wonder the software was terrible.
I remember being on site at a customer site some years ago, running some obscure DOS E-mail app which issued similar beeps which were difficult to disable.
This particular installation divided users up into many small groups of about 4 to 6 users, each group bearing the name or number of the cubicle the users were sat in. Every time a global E-mail went out it propagated around the office in alphabetical order by user group name - which meant a 10 second wave of beeps spreading linearly around the building from one end to the other.
I can remember a severe problem with Western Digital ISA Ethernet cards at one of our customer sites. There were at two completely different implementations of one board - WD8003E if I can recall it properly.
Later versions of WD's own official IPX and NDIS drivers would work with all card revisions. Our system, though, relied on monolithic drivers from the network vendor at the time, and our customer's entire set of new desktop PCs suddenly stopped working after a client software update as the newer drivers didn't work with their version of the hardware. As the client software had to match the version running on the server, the site had to downgrade all of their severs to get their desktop PCs back.
I'm sure something similar also happened with a popular SCSI adapter around the same time.
Same here with an obscure OS. Server's built-in NIC only supported once the operating system was patched, patch could only be applied from a client PC once server was connected to a LAN. The fix in this case was to connect the server to a WAN link via a serial adaptor, and then get someone at another site to login over the WAN and slowly patch the server.
We had similar issues with storage devices only being supported once patched. One fix here was to install a slower second supported storage adaptor, re-cable the drives, install, patch, and put the cabling back. Sometimes it worked, although disk numbers would be mixed up. The other was to simply wait for the vendor to eventually get round to issuing new release media with new drivers.
After a few years of this farce they eventually allowed hardware manufacturers to produce their own server device drivers, which were installed from floppy. Well, until the hardware vendors started bundling hardware utilities and diagnostic tools with their driver installations, causing most boot partitions to run out of space.
A testimony before a Parliamentary committee is not the same as an appearance in court - there could be legal ramifications for a broadcaster covering it if the details later turn out to not be true.
This is the same reason why the BBC and other broadcasters never cover major allegations made on the front page of national newspapers before the accused has had a chance to reply.
Ughhh!
Started using Lotus Notes before they were taken over by IBM, back when the only supported server platform was OS/2. It was used in-house for running various internal databases for the support, sales and admin departments, and we also did some development work for elsewhere.
My general experience was that it was fine for certain types of database which required just a few users, but bad for implementing E-mail; the latter confirmed by discovering the arcane process that ran nightly to purge attachments from deleted E-mails.
At one point my workplace developed an intranet produce that used the web-server component of Notes to serve documents internally, with a few third-party libraries thrown in for document conversion. The Notes web-server was singled threaded back then, and the search mechanism, whilst thorough, was somewhat slow, resulting in the server being potentially unavailable for minutes whilst handling a search query.
I can also recall other quirks, such as embedded Word attachments gaining the page dimensions of the form that they were embedded in when printed, which caused no end of fun with some printers.
I can recall Your Computer magazine printing BASIC listings for their own checksum utilities for several platforms.
Unfortunately at one point they cocked up this listing, so that after spending hours entering page after page of hex, the code would naturally crash as soon as you tried to run it, due to a subtle off-by-one error.
The only double-sided CD I can recall was the short lived DualDisc format, which was effectively a CD on one side and a DVD on the other. Unfortunately the resultant discs were to thick to adhere to the published CD specifications, and many CD players simply couldn't handle the CD side due to laser focusing issues. Some slot loading drives also had problems.
A double sided CD would be even thicker, and even more problematic.
I once had to use a 486 workstation that had a major design fault - the power on switch was on the front of the case and at just the right height so that pushing the keyboard back would hit the switch and power the thing down. The vendor of these things also touted them as low-end servers at the time. No wonder the manufacturer went bust.
As for floppies, I recall having to once install an early version of NT from a stack of floppies - over 40 of them - as the workplace had a policy of not ordering software on CD.
I used to work for a Yorkshire based IT company who had customers all over the country.
One week I was sent down to Sussex on a four-day course to cover an imminent new release of one of the systems we dealt with. On leaving the course on the final day to catch a train home, I was stopped at reception and told that there was a call from my office - this was before I had a mobile phone. Apparently one of our customers in the City of London had a problem with a server, and could I pop in to help out a hardware engineer who was having problems as he didn't understand our OS. Of course this meant that I would have problems getting home that evening, but I could stop at a salesman's house overnight.
So I made my way to the station, and caught a train to London Bridge, and then made my way to the customer's office for an evening appointment. The hardware engineer seemed to be pretty clueless, as all he had to diagnose faults were tools on a floppy that he didn't understand. He was convinced that the fault was with the motherboard, and he would arrange for one to arrive at the office for him to fit the following morning. Could I be on the phone the following morning to guide him?
So I headed off to call our salesman to arrange to stay the night in his spare room. I somehow managed to catch the last District Line train to his house to stay for a few hours, then on to catch the first train back to the office.
Tired and exhausted, I make it into the office the following morning, and start talking to this engineer. It takes him hours to get the system working again. Then once fixed, it's obvious that he has somehow managed to wipe the server's RAID array in the process, so could I guide him through the process of re-installing the OS and restoring a backup? By this time it is about 5pm in the evening, and I've only had one sandwich to eat all day, and most of the staff from our office have already left for the pub.
So I stay on the phone laboriously guiding him through the various steps to reinstall the operating system, which weren't that easy. The process was made a bit worse as we needed to install various patches to the server before we could attempt to recover from backup, and each one of these took some time. I remained in the office until about 10:20 that evening, caught the bus home, and then stayed on the phone to him until about 1am the following morning.
As I'd been away from home for a week on the course, I had no food in, and by 1am there were not even any takeaways still open, even on a Friday. So I went to bed starving, and missed a local event I had planned on attending.
I used to come across precisely the same thing when working in the support team for one of the handful of UK suppliers of an obscure system. Several times an incompetent support contact would get dismissed from one customer, only to turn up at another a few months later.
One was a particularly irritating person who couldn't do anything right, and ended up at three different customers of ours. One call alone, pertaining to him not being able to get our system installed on a new server due to a hang on booting, took up all of the day, due to him being totally unable to tell me the last line displayed on screen when the system hung. I forget what the actual problem was, but it would have been something to do with an incorrectly configured device driver.
Another salesman one. I received an internal call one day from a salesman that his Windows 3.1 E-mail client was running slowly, and could I sort it out.
So I wandered down to the sales team once free, and noticed him on the phone whilst reading through his latest E-mails. When he had finished reading a message, instead of hitting 'Close' or 'Delete' to dispose of it, he was hitting the minimise button, and he seemed to have over 50 different E-mail messages currently minimised. As this E-mail package displayed something like "Message 4 of 184" in the title bar of every message, every single minimised window had to be updated whenever a new message arrived.
The salesman's main task at this point was to sell this E-mail package to customers.
I used to run a helpdesk for external customers, but occasionally had to also offer support to our in-house team.
One particular head salesman was especially dim, and although he could remember his password, he repeatedly couldn't remember his user ID for the main network. As we'd added the sales group details to the default search list for his department, this was simply his first name and surname with a space in between, although we had kindly also added an alias to allow him to login using just his first name as there was no-one else in the sales team with that name.
So, in other words, he couldn't remember his own name.
There used to be a vinegar pipeline in Birmingham which ran between two HP sauce factories. To do so it had to cross over the Aston Expressway - i.e. the A38(M) - and was hidden inside one of the overhead gantries carrying matrix signs.
See here for a photo - http://pathetic.org.uk/current/a38m/photos/pages/000_0007_jpg.shtml
I worked with SCO Unix (and Xenix) back in those days too.
Every few months we put on a training course for the OS we sold, and one module involved setting up an SMTP server to communicate with an external system, in our case the office's hack SCO box we used for generic support.
Despite our best efforts, the antiquated version of sendmail supplied by SCO Unix simply refused to work properly, and most E-mails from that part of the course got lost.
So for one course I downloaded Slackware Linux and whacked it onto one of the training rooms spare DOS PCs using UMSDOS. When used on the course it worked flawlessly.
Back in the early 1990s I was involved in supporting an obscure RDBMS. Application code ran on MS-DOS PCs which communicated with a central server, which at the time seemed to be the only one which ran on our cranking server operating system. The resultant code was fairly huge, particularly if you elected to use their optional DOS text menu API.
As code was compiled on MS-DOS systems the various development files had 8.3 file names with fairly sensible names. The standard include files for the various APIs all had 6 letter file-names - the SQL API was SQLAPI.H, and SQLAPI.LIB for example.
However for development testing purposes they also provided a standalone server that ran in the background on a development PC. To compile against this meant substituting some of the API files with their standalone server variants, which meant adding the letters SS to the end of each include and library file - hence SQLAPISS.H and SQLAPISS.LIB.
Spotted this in the Java source code of a tool I used to use to set up new user environments on a financial test system:
After the username and password had been supplied, there was supposedly a test to ensure that the password was not the simply the same as the username, with or without a case change. However the numpty who had coded the thing was simply testing to see if the password matched these precise strings: 'username', 'USERNAME', 'UserName', 'Username', 'UsErNaMe' and one or two others.
Myself and a couple of colleagues once had a site visit between Christmas and New Year, upgrading and patching the obscure OS on a couple of dozen file servers, and then upgrading the client workstations. My servers upgraded fine, but my colleagues had serious problems getting the upgrade to take on other servers, caused by OS changing the order it allocated storage controllers between versions without documenting it. As our supplier's UK support centre was closed for the holidays, and with this being pre-internet days, getting a fix was tricky. Still, with some playing with EISA configs we got it working.
There were certainly 80186 processors fitted in some of the hardware I had to deal with 20 years ago.
I'm sure there was one in the original version of 6 port serial card which I had to deal with almost daily, a horrible ISA card that needed a 128K hole in the PC's memory map to function. Finding a suitable hole wasn't that easy, and was made worse when customers wanted two or even three of these in one machine. You could just about get them to work on EISA based systems, but not all of them. Anyway....
There may have also been one in a fax processor board we also used.
There was a rights management system name Copycode that some aspects of the recording industry were attempting to get implemented circa 1987, when DAT was being launched.
That system worked by cutting a notch into all recordings at around 3.8kHz. Recording devices fitted with Copycode circuitry would look for the absence of any signals at this frequency, and if so would prevent recording.
The people behind Copycode claimed it was inaudible. Many others disagreed, and the system was often mockingly called 'The A-flat Remover'. It naturally was never adopted, but the farce behind it was one of the contributing factors to DAT not being taken seriously as a home system.
Another reason for the need to keep Windows XP is that it was the last version of Windows that was able to communicate via NetBEUI, the extended NetBIOS over Ethernet protocol.
Although it died out as a standard peer-to-peer protocol for PCs many years ago, there were some automated manufacturing systems that used NetBEUI for control purposes, and some of these are still in use. The use of a protocol other than TCP/IP prevents such systems from being implemented in a VM.
Back in late 1999 I did some work at the head office of a major high street name who were in the process of setting up their own E-commerce site.
The password for the main NT domain was just 'password'. To make things worse, there were posters everywhere highlighting their commitment to security.
The Mosquito also acted as a freighter. Several Mosquitos in BOAC colours were used to transport ball bearings from Stockholm to Leuchars, using the bomb bay as a cargo store.
The bomb bay in these Mosquitos could also be used as an improvised cabin to transport a single passenger. The most notable passenger to be flown this way was physicist Niels Bohr who was smuggled out of Sweden in 1943.
Almost every customer of ours were more or less forced to move to another OS due to Y2K.
The weird OS we resold was very late at getting a Y2K compliant version out of the door, and when they did they dropped support for various bits of hardware and add-on software. One critical component was the vendor's range of custom communications cards, which were used for a whole range of purposes, including high speed asynchronous links, IBM mainframe connectivity and plain old dial-up modem connections. No real reason was given from dropping support for these cards, and Y2K simply didn't make any sense.
As the majority of sites used these cards for one purpose or another, they had to look for a workaround, and as workarounds were thin on the ground they dumped the system for something else - often Windows NT.
DIP switches were only part of the problem. Getting pound signs to print from DOS applications was the real challenge.
The majority of printers used language selection DIP switch to replace the hash sign with a pound sign, i.e. a character value of 35 decimal, but the DOS applications themselves, if using codepage 437, used 156 decimal for the pound sign. So you needed to set up a character translation table, assuming the application had one.
I had one application whose supplied drivers were strange - a UK one for my printer produced the pound sign, but it didn't allow me to print charts. The driver that did print charts didn't print the pound sign. As documents contained both text and charts, it made it impractical to switch drivers partway through a printout. I ended up reverse engineering the printer driver files (which were effectively just a table of escape codes with an optional translation table), and combining the two together to produce one that worked properly.
I suspect that the proposed 16-bit variant of the Sig-Net would have been earmarked to use Software 2000's TurboDOS. This was a multi-user operating system which borrowed a lot from CP/M, and which ran on the Intel 8086 series of processors.
I briefly supported a legacy TurboDOS system back in 1990. The main unit featured on 8086 processor, a hard drive, an S100 bus, a dumb terminal, and various printer ports. Into the S100 could be plugged a numbe of 'slave' boards - these each supported two extra users, via two 8086 processors and two dumb terminal ports.
Around the same time, an FTP server at company I worked for was similarly hacked. In this case, though, the hackers uploaded a large archive of cracked application software.
The FTP server was running on an old Solaris development box assigned to one project. I suspect that the hackers got in via an exploit, and not a leaked password.
There were several other problems with Banyan, other than their marketing department:
Beyond Mail, their enhanced Windows E-mail client they acquired, was dreadfully buggy. In particular version 3, the 32-bit Windows version, which constantly crapped out with error messages when handling rules that worked fine in the previous release.
Hardware issues: They were hampered by a limit of 2GB per file-system for far too long. They were also very late in releasing a means of allowing hardware vendors to create their own server device drivers for NICs and storage devices. Their own serial card, the only one supported for many server-to-server comms, was notoriously difficult to configure on many server platforms, until it was re-designed many years too late.
Then, when Windows NT Server was taking off, Banyan tried to support long filenames. They cocked up the handling of codepages badly, that many systems suffered serious corruption, at lest when using codepages from outside the US. That was the last straw for many sites.
Ah, yes, page after page of hex-dumps of machine code and/or data to POKE into RAM.
Your Computer was particularly prone to listings like this. Initially these were printed without any form of checking. Eventually they had the bright idea of including a simple check-sum after every 32 hex characters, and also produced a small BASIC program for each platform to check this checksum and POKE in the hex. Unfortunately the initial version of this program they listed for the ZX Spectrum, although appearing to run fine, didn't write the hex into memory properly due to a subtle off-by-one error!
This was a wretched machine in many ways. The standard machine had just 4K of RAM, and with half of that dedicated to its character mode display, there wasn't that much left to do anything useful. Furthermore it was saddled with a cut-down version of BASIC, lacking essential features such as FOR...NEXT loops.
So to get anything really useful out of the machine you needed to obtain both a memory expansion module and the Extended BASIC cartridge, and to use both simultaneously you also needed to acquire the expansion module. Once you'd shelled out the cash for all of those, it would have been far cheaper to have gone for something more featured in the first place.
Even the better quality games produced by Mattel were essentially conversions of existing Intellivison titles, the main aspect of the conversion process being to convert them to use the Aquarius's poorer graphics.
I can remember actually seeing a Sord M5 in the flesh, running in the newly established electronics section in the back of a nearby toyshop. The same shop also tried to sell the equally poor Mattel Aquarius.
There were, though, several other really appalling home systems that appeared around the same time that probably won't make such articles, due to the complete lack of UK sales, and really poor specifications.
One was the COMX-35, a Hong Kong produced machine with a couple of notable features. One was the choice of processor, an obscure RCA produced one that was commonly used in space hardware at the time, but was really too slow for a machine of this nature. The other was the really buggy in-house written BASIC interpreter, which would often hang instead of producing error messages.
Another was the Laser 200, again from Hong Kong. This looked like a cheap ZX Spectrum knock-off, but with even poorer graphical capabilities. It was somewhat successful in Australia, when sold under another name.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
