September 23rd, 2015

Hard Stare

BASIC: good times, bad times, how to lose big

(The title is a parody of )

Even today, people still rail against the horrors of BASIC, as per Edsger Dijkstra's famous comment about it brain-damaging beginner programmers beyond any hope of redemption:

I rather feel that this is due to perceptions of some of the really crap early 8-bit BASICs, and wouldn't have applied if students learned, say, BBC BASIC or one of the other better dialects.

For example, Commodore's pathetically-limited BASIC as supplied on the most successful home computer ever, the Commodore 64, in 1982. Despite its horrors, it's remembered fondly by many. There's even a modern FOSS re-implementation of it!

I've long been puzzled as to exactly why the Commodore 64 shipped with such a terrible, limited, primitive BASIC in its ROM: CBM BASIC 2.0, essentially the 6502 version of Microsoft's MS-BASIC. It wasn't done for space reasons -- the original Microsoft BASIC fitted into 4kB of ROM and a later version into 8kB:

Acorn's BBC BASIC (first released a year earlier, in 1981) was a vastly better dialect.

AFAIK all the ROMable versions of BBC BASIC (BASIC I to BASIC 4.62) fitted into a 16kB ROM, so in terms of space, it was doable.

IOW, CBM had enough room; the C64 kernal+BASIC were essentially those of the original PET, and fitted into an 8kB ROM, I think. And the C64 shipped after the B and P series machines, the CBM-II. OK, CBM BASIC 4 wasn’t much of an improvement, but it was better.

Looking back years later, and reading stuff like Cameron Kaiser’s “Secret Weapons of Commodore” site:

… it seems to me that Commodore management never really had much of an idea of what they were doing. Unlike companies such as Sinclair or Acorn, labouring for years over tiny numbers of finely-honed models, in the 8-bit era, Commodore had multiple teams designing dozens of models of all sorts of kit, often conflicting with one another, and just occasionally chose to ship certain products and kill others — sometimes early, sometimes when it was nearly ready and the packaging was being designed.

(Apple was similar, but at a smaller scale — e.g. the Apple /// competing with the later Apple ][ machines, and the Mac competing with the Lisa, and then the Apple ][GS competing with the Mac.)

There were lovely devices that might have thrived, such as the C65, which were killed.

There were weird, mostly inexplicable hacked-together things, such as the C128, a bastard of a C64, plus a slightly-upgraded C64, plus, of all things, a CP/M micro based around an entirely different an totally incompatible processor, so the C128 had two: a 6502 derivative and a Z80. Bizarre.

There were determined efforts to enhance product lines whose times were past, such as the CBM-II machines, an enhanced PET when the IBM PC was already taking over.

There were odd half-assed efforts to fix problems with released products, such as the C16 and Plus-4, which clearly showed that management didn’t understand their own successes: the C64 was an wildly-successful upgrade of the popular VIC-20, but rather than learn from that and do it again, Commodore did something totally different and incompatible, launched with some fanfare, and appeared mystified that it bombed.

It’s a very strange story of a very schizophrenic company.

And of course, rather than develop their own successor for the 16-bit era, they bought it in — the Lorraine, later the Amiga, a spiritual successor to the Atari 8-bit machines, which themselves were inspired kit for their time.

This leaving Atari in the lurch, but to which the company responded in an inspired way with the ST: an clever mixture of off-the-shelf parts -- PC-type where that was good enough (e.g. graphics controller), or from the previous generation of 8-bits (e.g. sound chip), plus a bought-in adapted OS (Digital Research's GEMDOS plus GEM, never crippled like the PC version was due to Apple's lawsuit, meaning PC disk formats and file compatibility. And of course the brilliant inclusion of MIDI ports, foreseeing an entire industry that was around the corner.

The ST is what the Sinclair QL should have been: a cheap, affordable, usable 16-bit computer. Whereas the poor doomed QL was Sinclair doing its trademark thing too far: a 16-bit machine cut down to the point that it was no better than a decent 8-bit machine.

Interesting times.

Whereas now, almost all the diversity is gone. Today, we just have generic x86 boxes and occasional weird little ARM things, and apart from some research or hobbyist toys, just 2 OS families -- Windows NT or some flavour of Unix.