Talk:Taligent

Cards

The cards still exist in the possession of a manager at Apple -- I saw them in 2001 ;-) -- 00:18, 24 March 2003 User:65.93.180.198

Mach, Speed

I was working at IBM between 1985 and 1998. In 94 I had dealings with Taligent and AIM, attending a programming boot camp at the Taligent office in Cupertino. As I remember IBM's WorkplaceOS and Mach architecture grew out of Taligent and not vice versa. It was a shame as the Taligent GUI was very nice, but at that time it was very, very slow as it was running on AIX.

Cheers, PBA (23:26, 23 May 2004 User:Pbadams)

The Mach (kernel) came from Carnegie Mellon. But don't blame AIX for slowness. I got to audit significant chunks of the Taligent code, and it was far and away the worst code I have ever seen, to this very day. They committed every OO-design crime know to man, and invented new, unimaginable ones. I remember reading the graphics code: every pixel was an object. Simply changing the color of a single pixel would cause a dozen constructors and destructors to run. Drawing a line, or performing window clipping, was an unmeasurable cascade of useless constructors and destructors. Compare to, for example, the Bresenham algorithm, which could draw lines with a handful of cpu instructions, or hardware of that era, which could draw nearly a million lines per second. Of course the performance was a disaster: it was designed to be so super-flexible, so uber-customizable, so over-archingly object-oriented, that it had lost all touch with reality or pragmatics. I recommended to my division to walk, no, run away from it, and so helped put one nail into its well-deserved coffin.

BTW, I think the code-quality issue should be mentioned in this article, however touchy it may be ... linas (talk) 22:51, 5 December 2008 (UTC)

I'm not sure which version of the code the above comment refers to, but as a co-developer of the graphics library from '89 to '94, the description ("every pixel was an object", "drawing a line ... was an unmeasurable cascade of useless constructors...") is completely false. The lead designer for much of the project also participated in the implementation of Apple's Color Quickdraw, and similar algorithms (such as Bresenham line drawing) were used. I can't vouch for what happened to the code after '94, but certainly up until that point the above description is not accurate. 19 September 2009

Hubris

I do remember the hubris of the Taligent folks, which led to a sarcastic comment the floated around the industry regarding the attitude of Taligent. "Hi! We're from Taligent. Give us the money." Considering the current state of the world's most prevalent OS, perhaps they were justified.

(10:37, 22 January 2006 User:205.243.153.121

Yeah, they *did* think they were a gift to mankind, didn't they? linas (talk) 22:51, 5 December 2008 (UTC)

Page needs work.

This article needs to be seperated into different headings (i.e, "Pink Cards", "AIM joint venture", "The fall of Taligent", etc. The article also seems too much like a one-sided conversation. Not encyclopedic enough. Otherwise, it's great.

-- While it might benefit from more subheads and possibly citations, I disagree that this article is "not encyclopedic enough." The writing isn't the least bit conversational, it has no personal pronouns, and it focuses primarily on the facts of the company's lifespan. Also, having worked at Taligent during pretty much its entire existence, I can vouch for the accuracy of the article. Badken 08:59, 22 May 2007 (UTC)

Name origin

I don’t actually know the origin of the name, but about ten years ago I heard it explained as “Talent without NT, Intelligent without Intel.” This would explain the strange-sounding name. --193.11.177.69 19:45, 8 July 2006 (UTC)

I worked at Taligent from '91 until the end. The name was created by a professional marketing agency. No explicit relationships with other brands should be inferred, it's just what some external group of name-brainstormers came up with. Badken 08:54, 22 May 2007 (UTC)

Ouch - Slammed by The Register

"You can read more about the (real) Pink here, here, and here. Though we suggest you avoid Wikipedia's Taligent entry, which is unsourced, riddled with mistakes, and could have been written by almost anyone." Common guys... step it up. ;-) Morphh ^(talk) 19:24, 04 October 2008 (UTC)

• It is easy to allege mistakes when you don't have to prove your assertions. As of this writing, I believe the Taligent article is largely accurate, though very sketchy. So much could be said about the company, its culture, the architecture of the system, the political intrigue between the corporate players, etc.
  
  FWIW, I am a former employee of IBM and Taligent who was involved with Taligent on both sides of the company divide. I wish a journalist would research the corporation and write a book on it. I'd love to be an anonymous source. ;-) Vantelimus (talk) 21:20, 4 October 2008 (UTC)

Workplace OS

An assertion was made in the first section of this talk page that Workplace OS grew out of Taligent. It would be interesting if someone can source information on this subject. By the time I got involved in the efforts on the IBM Microkernel (early 1992), Workplace OS was at least a primordial strategy. IBM had been experimenting with Mach (the mk68 or so release from CMU) for a short while under the aegis of Larry Loucks. IBM had a small group working on what would be called the IBM Microkernel. They were outsourcing some experimental development to people at CMU and at the OSF/RI. Looking at the timing of things, it could very well be that Larry got his idea for the Workplace OS from Pink -- specifically, a microkernel with different personality servers running on top, Pink and Mac/OS for Taligent; OS/2, DOS, and AIX for IBM. However, I doubt one could establish such a clean provenance. These ideas (microkernels and personalities) were floating around the OS community at the time. Larry was likely influenced by that and IBMs vast experience with virtual machine architectures. I think the best that might be established is that the ideas for Pink preceded the explicit idea for the Workplace OS. And that once Taligent had been formed, IBM sought to merge the efforts. IBM did push Taligent into using the IBM Microkernel as the base for Tal/OS ostensibly to allow the Pink personality to run side-by-side with the OS/2 personality. (No, I'm not naive enough to believe that the ostensible motive was the only motive or even a true motive. The only ones who can speak definitively on that are perhaps Larry, Jim Cannavino, or Lee Reiswig. Having been a party to the Microkernel work on both sides (at IBM and then at Taligent) to say I have my doubts about the true motives would be a vast understatement. Others' opinions may vary.) Vantelimus (talk) 21:52, 4 October 2008 (UTC)

Larry Loucks goes back farther than that. He was somehow a driving force behind the Virtual Resource Manager (VRM), a microkernel for the IBM 6150, circa 1986-1988. Although it was commerical, and worked quite well, (it ran AIXv2 and PickOS side-by-side), it was abandoned in AIXv3. He was miffed about this, and took several shots at redemption. The most promising (but doomed) effort was to have a microkernel for the then-new PowerPC chip: both OS/2 and AIX were supposed to run on it, (there was even some tie-in to SCO's OS too, somehow, maybe I'm mis-remembering). I think Larry was honest: he really thought that one microkernel would be enough, and you could build everything on top.

But it never worked: features were missing, it would crash. One of the better kernel programmers I know spent six months on an emergency rescue mission on it, and to this day is happy to complain on what a horrid mess the thing was. Sometimes, technology dies because its just plain bad. Funny how that is. Sometimes technology dies because the executives, programmers and architects, and researchers too, over-estimate their abilities, under-estimate the difficulty of kernel design, and hire on young journeymen to do the work of experienced masters. linas (talk) 23:10, 5 December 2008 (UTC)

I'm not sure what you might be misremembering about a tie-in to SCO. But, the microkernel group at IBM in 1992 and 1993 was talking to every company that had an OS (except for microsoft) about using the IBM Microkernel as its base. Besides being involved deeply in the Taligent relationship, I was also involved in meetings with Novell. Later in 1993 and 1994, as Taligent's rep to OSF/RIs Operating System Steering Committee, I saw Larry lobbying HP, DEC, and everyone else. No one was buying his story.

Larry tried to bully the TalOS group into becoming a device-driver writing subsidiary of his IBM Microkernel I/O team. It was very clear that IBM had overreached and desperately needed more bodies to help it accomplish what Larry was promising. IBM became even less cooperative about the microkernel at that time, which stimulated me to work on dropping the IBM microkernel and sourcing one directly from OSF/RI. Taligent decided to drop the idea of an operating system for other reasons in favor of a portable application framework (CommonPoint) before that trigger was pulled.

Yes, Larry did think one microkernel would be sufficient. I don't think that was too outrageous a belief at the time as it was held by many people spread around the industry. Whether that one microkernel would come from IBM was the sticking point. No one really relished the idea of being dependent on IBM for the guts of their operating system. My experience working with IBM on Microkernel issues from the Taligent side was a good indication of why. IBM was too interested in their own needs and pandering to the next prospective customer to ever deliver working code to Taligent on time. I remember one drop of code around mid 1993. IBM was to deliver posix-like monitors and conditions and semaphore support. When we got the microkernel, I compiled and tested the TalOS synchronization classes I had written and they just wouldn't work. Indeed, if there was any contention on a lock, the system would crash. When the bugs were reported, IBM explained to us that the they had met their obligation by delivering a working system. When I protested that synchronization didn't work, they explained that the system was "working, but not functional". When pointed out that crashing didn't seem very functional, they said "the synchronization primitives work, just not in the contention case". Clearly, IBM was having trouble handling all their microkernel commitments.

I think Apple's Mac OS/X has proven that the microkernel strategy can work. They successfully migrated to a microkernel-based unix and used a Mac Classic personality to support compatibility with OS/9. This was essentially the same idea as Taligent's Pink and Blue box approach and IBM's OS/2, DOS, and AIX personalities on top of Mach. Vantelimus (talk) 12:26, 21 September 2009 (UTC)