October 6, 2000

2:56 PM

Fat chance.Since when do these four companies have the ability to prevent anyone else from creating yet another version? (Also, conspicuous by its absence from this story is Corel.)
posted by Steven Den Beste (4 comments total)

I doubt they're going to try to stop anything, expect a commitee consisting of the turbolinux crowd to start regulating some new "certification" businesspeople can get exictied about. It could turn out to be a good idea.

"TurboLinux is A.R.Z. certfied for compability blah blah" stickers will be everywhere.

"We were interested in running Peanut on these older machines but they aren't ARZ, so we'll pick this brand instead."

Dont ask me what ARZ stands for I just made it up.

posted by skallas at 3:11 PM on October 6, 2000

Also, these kinds of industry-standard efforts have badly failed in the past. Anyone remember "BASIC"? There was one fundamental question which was never settled in the standard: does an array start with index 0 or index 1?

The problem was that about half of the extant commercial implementations used one and half the other. If your Basic began with 0, then accessing element 100 of a 100-element array caused an array bounds error, because your 100 elements were numbered 0-99.

Ultimately, later commercial implementors of BASIC used a kludge: if you declared an array of 100 elements, it allocated 101 of them, and started with 0. That way you still had an element 100, in case the code you were trying to run had been written using a version of BASIC where arrays started with 1. But that was not formally part of the spec, so far as I know; it was just a kludge that the developers came up with to maintain backward compatibility with both standards.

And that kind of thing happens all the time. These guys can pontificate about how good an idea standardization is, but when they get down to the nuts and bolts, it's going to mean that some of them are going to have to change their own implementations, possibly in ways which are damaging to them. And eventually someone will refuse. Eventually they'll find their equivalent of the array-index issue where deadlock ensues and no-one will give in. There are just too many things which have to be resolved; there's no way they'll succeed in getting them all.

Meanwhile, anyone who wants to can create their own version and rearrange things to their heart's content, violating the spec anyway.

I think the only way they could possible get everyone to agree would be to go to an existing official standard for Unix (e.g. AT&T's Unix V or SGI IRIX) and simply say "We'll do it that way", and then EVERYONE changes to match it no matter how much pain it causes. (And everyone's going to have to change, too.) Unfortunately, as soon as you do that, all sorts of installers and programs break and all hell breaks loose.

By the way, this sounds a great deal like they'll soon bless one of the three GUIs as "the official one". That should be interesting. (Seems like they'd have to, so that the ISPs know which one to code to.)

Not even Linus Torvalds is capable of imposing order on this chaos. It ain't gonna happen.

This is one of the cases where commercially developed software has a major advantage. Windows is whatever Microsoft says it is. No-one else's opinion matters. OSX is whatever Apple says it is. No-one else's opinion matters. To answer the question "What is standard OSX?" you just go to Apple, sign the NDA, and purchase about a 5-foot high stack of documentation. Then you, the ISP, go away and code to that standard. If OSX doesn't match that standard, it's a bug and you get Apple to fix it.

posted by Steven Den Beste at 3:13 PM on October 6, 2000

I didn't read this article as saying major Linux vendors were going to force the world to adopt a standard so much as saying a standard was inevitable as a natural consequence of the way things were going.

This is just going to be a constant fact of life in the free software world. New things will come out - next year we'll have three or four journalling filesystems to choose from. After a few years, the majority will have picked one, dedicated minorities will support the others, and we'll move on to other problems.

It doesn't mean that someday, we'll all be using One World Linux; it just means that features which are one distribution's competitive advantage today will be every distribution's standard tomorrow.

-Mars
posted by Mars Saxman at 6:25 PM on October 6, 2000

I dunno, Mars, speaking from the "corporate pulpit," I read it differently: "We know that big corporations aren't going to risk mission-critical on Linux until we can actually say what Linux is." Amongst my peers in the corporate IT world, I hear lots of discussion about how they fell or were pushed into the "you gotta see this Linux thing! It's gonna save the world!" - only to discover there isn't, in fact, a "Linux thing," there's as many Linuses as there were Unises and all of a sudden we're all right back where we started from. It's not going to be enough to get somebody to put an easy-to-use, graphically pleasing consistent face on Linux; somebody's gotta make it consistent in the back end, too...
posted by m.polo at 7:29 PM on October 6, 2000