Now the bright light's gone

March 28, 1997

"So who owns the Amiga these days, anyway? Does anyone still make 'em?" is a question I get asked a lot. People don't in general keep up with the news - oh, they follow the scandals, but more people on the street can tell you about Tonya Harding than can tell you whether Apple is still in business or not. I'm not talking just about technophobes and the mythical Joe Public, I'm including people who run computer companies for a living, who cannot tell you whether the maker of the single biggest-selling production model computer of all time - the Commodore 64 - is even in business anymore. It's all the more disheartening when you have to explain the complex answer to the question - the Annual Amiga Ownership Shuffle every April or so, the many players (one company manufactures 1200s, one makes 4000s, one's making PowerPC stuff, two more are writing new OSes, and one's just threatening to sue them all for trademark infringement), the fact that, although new Amigas are not currently being manufactured, it's just a recent thing as the last production run was only a few months ago, and QuikPak will make more when they get money, and on and on. In other words, who owns the Amiga and do they still make 'em, you cannot answer with a smile, and you often cannot answer simply or plainly.

But we finally have an answer to that question the next time someone asks.

And the funny part is? Look at your calendars. They jumped the gun and sold the Amiga four days early. The Annual Amiga Ownership Shuffle took place in March this year, instead of April.

The takers?

If you haven't already heard, I suggest you sit down. It's not easy news to take:

Gateway 2000.

Yes. All right, I confess, I'm having trouble with the concept myself. This is the Gateway 2000 who ships computers in boxes made to look like cows. This is the Gateway 2000 whose magazine ads are almost as annoying as Midwest Micro. This is the Gateway 2000 who ships computers with 127-key programmable keyboards.

Now let's do some thinking here for a moment and follow some thought tangents. I should have been doing some thinking on this instead of trying to prove it a hoax when I first heard it. But hey, the source said "there'll be a press conference next Tuesday" - and next Tuesday is April first. (More on April Joke's-On-You Day in awhile - needless to say, it's my least favorite holiday, and Amiga people seem to be victimized by it more so than other platform owners.)

At any rate, just yesterday I was trying to figure out which PC brand to recommend to someone who absolutely had to have a PC, no Macs or other real systems would do, it had to be a PC. Compaq's far too nonstandard - even the BIOS is accessed through disk-based software. (They were the first, and for many years the best, PC clone maker in the world - but that was a long time ago.) Packard Bell, well, when Apple announced they were licensing the Mac OS for clones, PB were the first to apply, and the first to be rejected - on the grounds that their systems were such poor quality, even Apple, Home of the Exploding Powerbook, didn't want to be associated with it. Even mighty IBM still occasionally thinks it owns the business - and thus will get a little bit nonstandard occasionally, and while Mac and Amiga and UNIX markets thrive on cool proprietary stuff, in the PC world, 100% compatible must mean 100% compatible, any less and Win95 starts getting unhappy. Hewlett-Packard is a bit too big and expensive. Acer are just run-of-the-mill clones with wild-looking cases. Which leaves... Gateway 2000. Good-to-excellent quality machines, coupled with THE BEST customer service in the business, as rated by every PC magazine on the planet, year after year.

Now think what this means. Amiga owners, can you remember a time when Commodore helped you with a problem? Mac owners, now that Apple's charging you for tech support on the SOS/Apple line, are you particularly proud of the customer service you get? Commodore used to ship Amigas without motherboards - or boxes without computers. Tech support? What's that? ESCOM, fer cryin out loud, had NOBODY in their organization who supported Amigas, except for one store worker in England and the employees of the Amiga Technologies "division" - most ESCOM store workers and phone people were Amiga-hostile. Farther back for those who remember, Tandy's "habit" of working only on unmodified computers, often returning a repaired system to you with all your modifications in a little plastic bag; you'd think Radio Shack would mean great customer support, the truth was far from it, there are many tales of systems coming back from Shack repair in worse condition than they entered, screws missing, mains reversed, etc. Thus: as of today, for the first time in my life, I own a computer for which I can get decent customer support. :-)

(That's not entirely true - this 1200's so heavily modified there's no technician, Amiga-qualified or otherwise, who'll touch it. Even I'm scared to work on it sometimes. Y'know, I never did get the internal floppy resurrected after adding the Dataflyer...)

Next thing to consider: Gateway's WebTV.

Now, I've harped occasionally on the Macintosh TV, a unique but unfortunately short-lived product from 1993; it was a Performa 550-like Macintosh with a one-piece black case, CD-ROM drive, Trinitron color monitor inbuilt, and a TV tuner card that could direct the picture to the monitor, scaled to full-screen and all. It ran on a 32MHz 68030, came with a black keyboard, and was just generally the snazziest-looking Mac since the original. (It's still one of an elite breed, that of computers with a case colored something other than white or beige.) At any rate, it was a full Macintosh, and the TV tuner card worked great. The only real problem I saw with it was that the tuner card had to scale the TV signal to fit the monitor - which meant, on that crystal-sharp Trinitron, every pixel was clearly defined and sharp-cornered. QuickTime movies looked real on its 16-bit graphics card; live TV feeds looked like QuickTime movies scaled to full-screen. Anyway, the product was something of a commercial failure - and I believe it's one of the rarest Mac models around, alongside the Centris 610/DOS, the Portable, and the Anniversary Mac. I never understood why it failed; I loved the concept, and always thought Commodore missed an opportunity.

Gateway 2000 agreed at some level. Destination is a Pentium 120 running Windows 95, built into a huge 35-inch monitor-television combo. It runs on a wireless keyboard, and has a cordless mouse-like thing you use to control the pointer. (Don't know whether you point it at the screen or not; don't know whether it's a trackball or not.) High point: it runs all PC software, and runs it on a giant monitor so you can read icon text from across the room. Low point: it cost $3500 when a "normal" Pentium 120 system would only have cost $1500.

Your brains are working already. I can tell.

Yes, I see it already. The Amiga Television: an 030-based AGA motherboard, with hard drive, 6MB RAM, and a 33.6 modem, built into a 25-inch color TV. It's perfect for the job - the circuitry to output a VGA image to television costs as much as an Amiga 1200 motherboard. Without the need for an expensive scan converter, or the need to turn the TV into a huge multiscan monitor, you can throw a $500 Amiga mainboard into a $300 Mitsubishi color television, add closed-caption feed into the Amiga easily, maybe even add a framegrabber for $120, and there's your set-top box. There's a company who's already tried this, but its name escapes me at the moment so it apparently wasn't the market splash we wanted. But at any rate, it's worth considering - it's certainly a sleeker idea than VIScorp's buzzword-laden ED. The set-top box ceases to be an entity - it's just a television with a CD-ROM drive in the front. Transparent technology - what I've been preaching for awhile anyway. None of this "stealth computer" crap either - it is what it is, by building it into a TV and telling them up front it's not a computer, people won't expect it to be a Pentium 200, and will be pleasantly surprised to find out it sometimes acts like one anyway.

Next thing to consider, while I'm rapid-firing: Commodore for years went too heavy on their PC business, losing money in that hypercompetitive market. ESCOM went bankrupt when the bottom fell out of the European PC market - which left us where we are now. There's not a particularly good track record for companies who make both Amigas and PCs - perhaps you've heard the one about how you can't serve two masters, you'll either hate the one and love the other, or love the one and despise the other? It will be interesting, to say the least, to see how Gateway manages to keep the two product lines from weighing each other down.

It will also be interesting to see how they advance the product so that they aren't competing with their own Wintel clone boxes. They either have to position the Amiga as a high-end alternative to Windows - which means at least a year's worth of heavy R&D to make the sentence "high-end Amiga" sound less like an oxymoron - or the low-end alternative to Windows, which means they're competing with the PlayStation and N64, which will kick the Amiga's measly overpriced 32-bit ass and never look back. (That's a picture - imagine the various fighters from Tekken 2 teaming up with Mario to just beat the living shit out of a poor defenseless A1200.) The Amiga OS holds its own on similar hardware versus any other system on the planet - but unfortunately it's currently tied to hardware that's 6 years old. (Holy shit, is it 1997 already?) It can compete with Windows, if you can get it on hardware specced high enough; Gateway has a vested interest in NOT doing that, as it would cost a lot and they'd be competing with themselves in the end.

Gateway has virtually no European market share, despite their incredible sales here in the US. There's a possibility they'll start flooding the European Amiga distribution channels with their own PCs. There's a somewhat prettier but less likely possibility they'll just leave the Amiga network in place and use THAT as their entire European product line - sell Amigas to people who want them, and then they don't have to overreach either in Europe (to introduce their PCs) or in North America (to introduce their Amigas).

They have announced, however, that Amiga Technologies will remain a mostly independent division of the company - much as it was under ESCOM, but without the parent company crumbling on top of it - and will be renamed Amiga International. Who will lead this division? That's the good news - our friend Petro Tysch... Tychtent... Tyshenko... you know who I'm talking about.

On the surface it looks like the following good things will happen if Gateway does nothing else: With the Amiga assets out of bankruptcy court, they can start doing development again, hire personnel, and spring for another production run. The politics of dropping Gateway's name should make it just a teensy bit easier to get Amigas and software sold in department stores. Having the corporate finances report to North Dakota will be a tiny bit of incentive for Amiga International to do an NTSC run of 1200s. And best of all, as I said before, we no longer have to mumble and hide when people say "Amiga's out of business, isn't it?"

If Gateway decides to actually help the Amiga along, by infusing it with some cash, we should see new hardware designs (hopefully no more stealth turtles), more and bigger production runs (right now there is STILL a four year backlog of A1200s in North America), and maybe, just maybe, they can spring for dinky one-page ads in magazines to remind the world the Amiga still exists. How Gateway will be structured to make money off the Amiga isn't clear yet; it doesn't entirely make sense for them to pump money into a dying platform just for the sake of generosity, they want something from it, and hopefully it's not the Amiga nameplate on the front of a Pentium.

Wowza. At any rate, this has just been the capper on a truly bizarre month for me.

Couple weeks ago, as you know, Apple announced their big surgery: 4100 employees gone, and much of their high-tech projects thrown in the compacter. OpenDoc and Open Transport are gone, but Mike Markkula and Ellen Hancock are still there. The only REAL good news about the cuts, other than QuickDraw GX's long-expected demise, is that the Newton survived. Apple's trying REAL hard to cut that loose, though, possibly to spin Newton off into its own company, so it doesn't collapse if Apple does, and neither does Apple hurt when the Newtons don't sell. (The Newtons stayed, of course, because of the MessagePad 2000, based on a 162MHz [!] StrongARM processor but still running on batteries that last as long as on any other Newton. With technology like THAT coming from the labs, I think a few people found it too bitter to swallow, so they left it alone.)

What's out? Well, OpenDoc was an object-oriented component architecture. OpenDoc "parts" (applications) live inside OpenDoc "containers" (other applications) - thus a spreadsheet part could live inside a word processor document, have a Web browser in there, a 3-D rotating model, it's all very document-centric. At any rate, it didn't take off as Apple planned, partly Apple's fault for not marketing it, but it's gone now.

Also gone is Pippin, Apple's answer to the CD32. Pippin's a stripped-down Power Mac designed for set-top applications; as a Mac it's limited, as a set-top it's too expensive. Neat idea, but the Mac is not the computer to do it with, until they shrink the OS somewhat. Apple will still license Pippin technology to anyone who wants it - but will no longer develop it. Good, I say.

Open Transport is perhaps the most mourned death of the whole bunch. Originally planned to be Copland's networking mechanism, OT was released as beta last year, while Apple fine-tuned it and applications developers got used to writing for it. Now, they've decided that while Rhapsody will support OT in its emulation layer, they will not develop it further. It's all that much worse because certain critical bug fixes and features will never happen - from the oft-mentioned Ping of Death bug, to features like multiple IP addresses (which would make the Macintosh a SERIOUS contender in the Web server world) - it'll never happen. In its place, inside Rhapsody's native layers, will sit a Berkeley Sockets interface. This is the same sockets layer you'll find inside AmiTCP or Miami or AS/225. Many programmers think OT was the better technology; who cares in the long run, if Apple's just going to throw away all their technology every time they run short of cash. Whether it was cool or not, that Apple waited until it was 85% done and then dropped it, hints that Apple's real problems remain unfixed.

I've never been afraid of Apple going out of business... until now. Rhapsody is a concept, and a rough one at that. NeXT's high point was its user interface; the Mac's user interface has fallen behind the times somewhat. So what are they gonna do? Stick a Mac face on NeXT's foundation. NeXTSTEP, btw, is built on top of Berkeley UNIX, on top of a Mach microkernal. This will end up being UNIX with a Mac-like window manager. Existing Mac apps? Oh, they run in emulation modes inside the various colored boxes. There's the emulation layer, and the native stuff outside that; many Mac developers would like an intermediate stage, so they can tweak System 7 apps to use NeXT features, but no one at Apple has said they'll do it. No one at Apple has said they'll support 601-based Power Macs anymore, either; your 8100 is obsolete now.

Copland in its 1996 incomplete state, plus a year of development, would have cost far less than $400 million to complete. Instead, Apple spent $400 million on a company most pundits said wasn't worth quite that much - and now has to double-time it to put the Macintosh onto a UNIX base in time to deliver product sometime in early 1998. Thus Copland - in its third or fourth incarnation - will wind up only five years late when it finally reaches your Mac this time next year. Be, for all its imperfections and rough edges, is a preemptive OS solution now, and will run Power Mac apps long before Rhapsody does. (It already runs 68K Mac apps - one of two ways, either a modified Mac Plus-like Shapeshifter running inside UAE, or through a third-party 68K Mac emulator that's still in early beta. Power Mac stuff will be trickier - I think the Power Mac lacks the context-emulating abilities of the 68000 series, making Shapeshifter difficult to impossible.)

Apple needs to learn to finish what it starts. Projects that are so far behind they're hurting the company should have been throttled back long before they got to that point. NeXT/Rhapsody seems to me an impulse purchase - "we need an OS, here's a good one." I wonder if the purchase was influenced by Steve Jobs' legendary reality distortion field - "spend $400 million on an OS it will take a year and a half to adapt to what we need? And put you back on the payroll? Great idea, Steve!" I'd hoped Steve's presence would rejuvenate Apple's artistic soul - remember the difference between Steve Jobs and Bill Gates is that Bill Gates will ship an imperfect product and never blink - but it appears that isn't happening. Rhapsody isn't going to be the Copland/Gershwin of our dreams - it'll be UNIX with a happy Mac on startup.

But we'll wait and see.

Kaleidescope 1.5 has been released, and its file format is well-documented. Kaleidescope, for those who don't use Macs or haven't heard of it, lets you replace the Macintosh window style with various artsy "color schemes" - and it's quite flexible. It's the descendant of the Aaron extension, that makes your Mac look like Copland (as seen on various TV shows and magazine screenshots) - and includes a Copland appearance. But it also can do the BeBox convincingly - yellow tab and all. It works with SmartScrolls, for proportional scrollbars. So I put Kaleidescope on the Mac at work, broke out the copy of ResEdit, and did some tinkering. Here's what I wound up with.

No, it's not exactly like Kickstart 3.0. Due to limitations in Kaleidescope, I couldn't make the window gadgetry extend to the window edges - so I wound up with this instead. A few of you with early A3000s may recognize this scheme as being eerily like the developer versions of Kickstart 1.4. What you don't see here are the system buttons, which are rectangular and recess and turn blue when you click them - just like the Amiga. I had a "Gadtools" version of the buttons for awhile, that had the left and right button edges two pixels wide for the added authenticity. The scrollbars are proportional (with SmartScrolls installed) and even light up in white like real Workbench 3.0 scrollbars - it's frighteningly authentic in that regard at least. There are several other aestetic differences I wasn't able to fix - mostly objects drawing over each other, incomplete graphics (it's actually the Copland scheme hacked all to hell, and in monochrome the Copland gadgetry is still sometimes visible) and some weird rendering bugs. In short, it's not ready for release yet - and besides, I'll feel like a traitor if I upload a file to InfoMac before uploading a file to Aminet. But when it's ready for human consumption, it'll be here first.

April First is approaching quickly, and it is my least favorite holiday, as I said before. This year, they're getting started early - last week's "Amiga with 8 Megs of Chip RAM" threads on USENET culminated with Ralph "The Guru" Babel dropping hints of a "vchipmem" program he would be uploading to Aminet... next Tuesday. Long about April 1 I'm so paranoid I can't get anything done - and people still blindside me. I hate it.

But at any rate, I long ago had an idea for a product I called "Octavius" - an MMU for your Amiga's custom chips. Physically installing it would be a nightmare - I needed to do a lot more thinking along those lines - but what it would do is provide a paged memory management system for the chipset, allowing 8 megs of chip RAM (which would reside in a SIMM socket on the Octavius board) to be paged in and out of the 2MB chip space. Obviously this would require heavy software patches. Obviously it'd be a pain in the ass to install on AGA machines - the chips are soldered down, and I doubt you could pull those memory lines out to the expansion bus. The only way to really do it would be to pull the chips and install Octavius under them - much like the DKB Megachip - which means contending with surface-mount stuff and the low-flying RF shield in the 1200. Seemed like a great idea at the time, but...

Of course, if this were possible, and could be integrated into the machine with any level of functionality, then Ralph would have ALREADY written something like vchipmem and sold it for $100 a pop. The trick here - as with the joke - is not the OS communicating with the chip, it's the chip communicating with memory. Nothing software-wise is limiting Agnus or the AA equivalent (her name escapes me, just shoot me) from going to 8MB or beyond - it's the chip's actual memory bus, the same reason a 68000 cannot hold more than 16 megs of any kind of memory, the same reason a Z80 couldn't do more than 64K. Virtual memory trickery has to live between the chip and RAM, and since the innards of the chip aren't designed to understand why the middle of RAM suddenly changed, and they can't be software-patched to understand virtual memory (the way the processor can), the concept is impossible, short of a chipset revision.

The alternative, though, is much nicer, and I wish someone would do it: rewrite graphics.library to use Fast RAM whenever possible. AGA machines with fast RAM also generally have a faster CPU; my 50MHz 68030 can push pixels much faster than the blitter. Pushing those pixels around fast RAM, using it for the offscreen buffering (which currently happens in chip RAM, so the blitter can grab it when it needs it - this is why Amiga web browsers eat all your chip RAM so quickly, datatype-converted bitmaps are buffered in chip RAM until called for!) and then CPU-blitting them into chip RAM where the hardware can play with it (as needed) would go much faster than the current approach; doing most of the pixel manipulation in fast RAM would avoid chip slowdown (actually it's bus contention, EXACTLY the same problem faced by big multiprocessor computers! - which makes sense, as the blitter IS a processor as far as the Amiga is concerned) and would, of course, conserve the Chip RAM that's vitally needed for actually generating the display. Using the processor to directly blit pre-layered images onto the screen MIGHT eliminate some of the flicker you see in Web browsers and Workbench animation players - where not all the bitplanes go in at once. A cleverly written refresh routine would blit one byte to each bitplane, then the next byte, then the next byte - whether this would be faster than the blitter depends on the processor speed, but it would look a lot cleaner graphically.

The best solution, if you have a big box, is to stick a video card in there and use Chip RAM only for sound samples and non-card-generated screens.

It has been a strange, odd, bizarre, peculiar, quirky, weird month - and with three days left in it, things don't appear to be slowing down too much. Donna Lewis is singing Coke commercials now. Thirty-nine Web developers in San Diego killed themselves so the UFO following the Hale-Bopp comet (there is no UFO, that story was itself a hoax) could pick them up and take them home. Microsoft is launching a whole fleet of geosynchronous communications satellites. Apple's cut off their future to spite their face. My car's got a bad steering box. I have $40 in the bank. My ex-girlfriend has started drinking. And, of course, the Great Silicon Fruitcake has changed hands again. I WILL have gray hair before this is over.

Now, the big question is, do we go stick Boing balls and checkmarks to every Gateway2000 PC we find, or just paint our Amigas with black and white cow spots?