What should we do about [OS/2] PM vs. Windows? [1990] [pdf] | Hacker News Reader

What should we do about [OS/2] PM vs. Windows? [1990] [pdf]

groklaw.net

32 pointsyuhong14 comments

In 1990 the biggest competitive threat to what would become the most important single business unit in all of software was... Sun/SPARC. The biggest strategic initiative in that BU was... port to RISC.

I remember those days. I was doing some development under OS/2 1.3 in college and got my first job based on my OS/2 experience, hard to believe now.

When OS/2 fizzled out and we focussed more on Windows it was a rude awakening. The OS/2 API was very clean if a bit bureaucratic. The Windows API had a very disjoint and ad-hoc feel to it and of course the "OS" itself was awful. I do remember enjoying writing VxDs, though, that was fun.

Like DOS, Windows had one of those APIs that was written by someone who seemed too drunk to care about details.

Curious quote, "When will Scott McNealy realise that OSF is the best thing that ever happened to him and cease offering it [sic] technology?"

What was unique or interesting about OSF? There's quite limited resources online.

Sun was an exciting company with its own vertical technology stack, and I was always surprised that they never made a serious attempt at making Solaris the developer workstation of choice. To do this they would have needed to (1) maintain a first class web browser environment; (2) provide a less sucky window manager; (3) multi-monitor support; (4) something that handled PDFs nicely; (5) an adequate non-console email client. Something like Java but with a better GUI layer would have been nice too. They bought the Lighthouse guys but then didn't use them.

Solaris had a significant technology advantage over NeXT, an established base in enterprise (a lucrative marketplace and much more forgiving than home desktop).

Ironically, while Sun were busy burning their OpenStep stack, NeXT were embracing Java and that kind of saved NeXT. It allowed independent NeXT developers to ride forward on a maintained technology stack (WO/Java) while Apple retooled.

Would be interested to know anything at all about OSF, and what distinguished it from SVR5.

I don't think that OSF had anything special at all. At the time, it looked like HP and DEC (especially DEC) trying to keep Sun and AT&T from steamrollering everything else. Bear in mind that in the mid-to-late 90s, there were a bunch of commercial Unix variants (HP-UX, Mips, SGI, DG-UX, Convergent, Pyramid, Dynix, DEC, Mt Xinu, etc etc). Each of them was holding on by their fingernails, more or less. The OSF was self-appointed to anoint the successor to BSD that was supposed to become The Unix next, as opposed to letting AT&T and Sun figure that out. The more cynical also believed that DEC was sowing chaos to give itself more time to either conquer the world with VMS, or figure out what to follow-up VMS with.

PDFs did not even exist at that time, postscript was the document format of choice, you are also placing undue importance on web browsing, it really didn't explode till windows 95 came out.

> (1) maintain a first class web browser environment;

The web didn't exist at the time this memo was written. Microsoft only noticed it existed about 6 years after that.

> (2) provide a less sucky window manager;

Less sucky than what? Windows 3? OS/2 v1? Motif WM? twm?

> (3) multi-monitor support;

Mostly nobody had dual-head workstations at that time. Most PCs had 14-inch SVGA screens. 1024x768 8-bit color was a luxury.

> (4) something that handled PDFs nicely;

PDFs didn't exist until 93. Sun had excellent Display PostScript support.

> (5) an adequate non-console email client.

I liked the one in OpenWindows.

> Solaris had a significant technology advantage over NeXT

Sun had faster hardware. NeXT was much nicer to use.

> NeXT were embracing Java and that kind of saved NeXT.

Hardly. NeXT was doomed. Apple's "reverse acquisition" saved NeXT.

NeXT was based on Objective-C. No Java there.

webobjects was a rebranding of openstep + some new tools. The last version lived on java.

Some people started with next, branched to wo, survived on it for six or seven years and then migrated back to cocoa/iphone. Similar API through

Not directly in response to you but a couple of others - I was rambling on a diversion, and when doing that need to explain the new point more clearly than I did. Sun seemed to have been well-positioned for building a good developer desktop platform for a window of about fifteen years but seemed to never attempt it. The fact that next bounced their ecosystem off Java while they were building a platform partly demonstrates the strong position Sun had.

Notice for one thing that the limitations of these "extender techniques" were not even considered, such as no memory protection or separate address spaces.

For those scratching their heads (as I was initially), this seems to be a random memo from the people that worked/architected/strategized/marketed OS/2 and/or Windows, within Microsoft.

They are discussing how to better compare and contrast OS/2 and Windows, to their own (OS/2) advantage.

I'm not sure why this is an important memo, or what significance it has?

Look up "MS OS/2 2.0 SDK". Also see this: http://news.ycombinator.com/item?id=3419551

    To: Bobmu, Markz, Steveb, Nathanm, Johnsa
    From: Paul Maritz

BobMu = Bob Muglia. Currently SVP of Software at Juniper Networks. As of last summer was President at Microsoft for the Server & Tools Business. At the time he led program management for what would be come Windows NT.

MarkZ = Mark Zbikowski. Key developer on Windows. The "MZ" in the Windows portable executable format header are his initials.

SteveB = Duh.

NathanM = Nathan Myrvold.

Paul Maritz = Now CEO of VMWare.

Mark Zbikowski was one of the devs of DOS 2.0 and 3.0.

Reply on news.ycombinator.com