I keep wondering when I'll draw a final line under the extremely small software outfit Ansible Information, and retire to my other life of overwork as a writer. Certainly Ansible is beginning to flag, but -- like the residue of a lingering head cold -- it refuses to go entirely away.
2001 saw further decline when the immemorial Amstrad PCW 8512 lost the use of its B drive (no, changing the belt didn't help). A few months later, the A drive went too (ditto). This happened exactly as I was trying to move stuff from a pile of old CF2 disks along a cable to the PC.
The result was a solemn pact between Ansible and SD Microsystems, whereby the great Steve Denson copied that particular batch of disks and in return will receive all future Ansible disk-transfer enquiries. Another fragment of the business empire successfully dismantled!
Further dismantling of a cruel and unusual nature was later carried out on the failed PCW plus a couple of dead ones that had accumulated in our cellar over the years. From the fragments, I managed to assemble a patchwork system with an evil, gaping hole where the B drive used to be. It's like owning Victor Frankenstein's home computer, born of unhallowed parts stolen from graveyards, and apt to turn at any moment on its hubristic creator. There's the terrible suspicion that Igor may have messed things up and brought me an abnormal processor chip....
Still, the undead PCW does actually work. For now.
Meanwhile, Microsoft continued its policy of causing as much irritation as possible to mere customers. Practically the only Ansible product that still sells is Ailink, the package that reads your 3.5" PCW disks in a PC disk drive and then -- this is the cunning part -- converts the Locoscript documents to Rich Text Format files suitable for Word and other modern Windows word processors.
I have a suspicion that Bill Gates disapproves of this. At any rate, each version of the accursed Windows makes it harder for our software to read PCW CP/M disks on the PC. Back in the days of Windows 3.0 to 3.11 there was no problem. Windows 95 brought some difficulties: the CP/M copier needs to access disks in a special and nonstandard way which was now discouraged. I programmed around this snag, but it got even worse in Windows 98.
Revelation! If you restart Windows 95 or Windows 98 in MS-DOS mode (an option on the Shut Down menu), all these difficulties with the CP/M Copier go away. Naturally Bill Gates got to hear of this, and came up with his cunning counterstroke. The new Windows ME can't be restarted in DOS mode.
Thanks to customer feedback, input from that nice Mr Denson, and a little lateral thinking, I came up with an all-new Ailink workaround that was far less drastic than restarting the computer, and seems to do the trick on Windows 95, 98 and ME.
It was at this point that the evil Gates, laughing satanically at the futile struggles of mere mortals, released Windows XP. Maybe it would be a cunning plan to go out of business right now.
Meanwhile, although I'd vowed to have nothing to do with the non-PCW-compatible Amstrad PcW16, the inevitable finally happened in 2001. Someone with one of those little machines bought Ailink without heeding our subtly worded disclaimer ("If by any chance you need to convert PcW16 documents, PLEASE GO AWAY"). I was faced with the grim choice of bodging the PcW16 file format into Windows-readable shape, or offering a refund.
Naturally I chose the more complicated and time-consuming option. Just as Ansible's fabled software for the Amstrad PCW was written in Borland Turbo Pascal 3.0 for CP/M, its distant descendant Borland Delphi handles our current Windows 95+ programming. It was time to wield the mighty sledgehammer of Delphi upon the small and wizened nut of PcW16 document format....
Skipping over many tedious hours, I can report that PcW16 documents have a less utterly cryptic internal structure than Locoscript ones, although there are oddities that I haven't fathomed. The main obstacle to decoding them is what seems to be a bizarre attempt to save space by leaving out all the spaces (with mysterious exceptions). Instead, the ASCII coding of the first letter in each word (with the same mysterious exceptions) is increased by 127 as an indication that a space should be inserted just before. Very odd.
After various tweaks, the ad-hoc conversion to Windows Rich Text Format was producing recognizable prose and eliminating most of the formatting boilerplate. The bumf at the head of each PcW16 document varies in length: it was easier to delete by hand to the obvious start of the text than to automate this bit. A few inexplicable random letters still sprinkled the text, always at the beginnings of words, but these were obvious enough to the editorial eye.
So, another refund saved! It's not a saleable conversion program, though, because the results are scrappy and need hand-editing. All I need now is to discover that the full PcW16 document structure is publicly available on some website, and that other hands have already written a freeware conversion utility producing RTF results of unsurpassable perfection. This would provide suitable dramatic irony to conclude another episode in the decline and fall of Ansible Information.