Hacker News Reader

Comment by madaxe_again | Hacker News Reader

Something like tradition. The IBM PC has been an "open" system since the late 70's. Tablets etc (palmtops!) started appearing in the 90's, and were walled gardens from the get go.

View on HN

That was the whole thing about PCs - if my memory of Cringley's "Triumph of the Nerds" serves, the genesis of this category was IBM rushing to a market where other "closed" vendors such as Apple were already established. It wasn't that PCs were open "by design" but more that they did things quickly and gathered other vendors "off the shelf" notably Microsoft and Intel, and had a copyrighted "BIOS" holding the whole thing together. Once Compaq reverse engineered this BIOS (through some legally clean techniques) the market was blown open to the clone manufacturers and Intel and MS were happy to play along since at the end of the day they were just shipping more units .... probably couldn't do it these days due to DMCA and other modern copyright legislation in other jurisdictions.

And what Compaq did was possible because at that time you couldn't copyright the BIOS API. Today it might be different, to the loss of everyone.

We're still waiting to hear what's acceptable in terms of "fair use" but yeah I guess reverse-engineering to create a competing clone probably wouldn't be covered :)

IBM PC was made open by Compaq in 1981 not late 70's, IBM would rather had it closed as the Commodore, Spectrum, Atari, Amiga, Acorn, Mac were.

You're both wrong. The IBM PC was made by IBM in 1981 and it was absolutely not open. (Just hurriedly implemented using commodity parts so, other than the BIOS, it was pretty easy and legal to copy most of it) Then Compaq came along in 1982 and sold a clone of it... and in some ways it was even less open than the IBM. Neither one of those companies had any interest in the PC being 'open'.

It was the 3rd party suppliers (i.e. BIOS by companies like Phoenix, Award, AMI along with motherboard manufacturers etc.) selling picks and shovels to the would-be prospectors (i.e. clone assemblers) that allowed for the 'open' PC standard. For example, there was a company call PCs Limited that licensed AMI BIOS so that they make some IBM clones... they did kind of OK and soon renamed themselves Dell...

Would it have happened if Compaq hadn't done a clean room reverse engineering?

I think so. Compaq just proved that it could be done without being bankrupted by an IBM lawsuit. IBM was the 800lb gorilla like Microsoft was in the 90's or Apple/Google are today. Being sued into oblivion was the real fear, not that it was technically a difficult job.

Some of the other companies I mentioned were the ones that followed Compaq's lead and did their own clean room versions which they then sold to everyone else who came flooding into the market.

As an aside, this was actually a problem for the first few years: take multiple independent clean room implementations, bake in some silly things that the manufacturers did with them (gotta customize, right?), and combine it with the bad programming practices of the day (i.e. programs written in assembly doing jumps into undocumented portions of the BIOS because, hey, it worked on the IBM and saved a few bytes/cycles or did something useful...) It was similar to how well web standards worked not too many years ago with multiple browser implementations doing/interpreting things slightly differently and developers doing things they shouldn't because they could... and that was with a public spec designed to be copied. Eventually things get worked out, but both had a rough start.

Just a nitpick - Spectrum was pretty open (at least if we are talking about 48/128). My father bought us Spectrum specifically because it was more open than alternatives (Atari or Commodore), and he later bought the PC for the same reason (and I would never buy a Mac for the same reason).

Unlike its competitors, Spectrum had standard phone jack connectors for sound and magnetic tape recorder (so you could use any recorder). It also had a direct circuit board connection to the system bus on the back, and the schematics were available. My father built us the 8255 interface to have joystick and other peripherals connected.

Not if we are talking about installing alternative OS into it.

Actually, that was pretty doable too. Later, my father bought D40, which was a 5.25 inch disk unit produced by Czech company. The D40 had its own 16kB ROM which got mapped instead of the standard Spectrum ROM when NMI was invoked (the D40 ROM contained code to handle the disks). So it was totally possible for an external device to remap original ROM with something else, even RAM.

Regarding the RAM remapping: There was Didaktik Gama (made by the same company, that made D40), that had 80kB RAM. It was done by switching the 32768-65535 address range between two banks.

There wasn't an OS to talk about. The 16kB ROM contained a BASIC interpreter, which was ignored by most applications you would load from the tape. If you wanted to talk to the hardware (joystick, printer), you did the I/O yourself.

The 16kB ROM BASIC interpreter was the OS!

And you surely could not overwrite it.

Not from inside; but an external device could remap the address range (not that it was made often, except for D40/D80 which did that, replacing microdrive code with a floppy one) and the ZX128, which actually had two ROM banks.

There weren't many useful things in the ROM, if you were writing your application. Maybe the font table and tape loader. On the other hand, if you were able to ignore the ROM, you got few bytes more of RAM, because you were able to repurpose the "system variables" area.

The biggest "problem" with ROM that you were unable to replace it with RAM (until next reset, for example), thus taking precious 16kB out of 64kB address range.

Reply on news.ycombinator.com