Talk:Xerox Star

From Wikipedia, the free encyclopedia

Contents 1 Why Star was slow 2 Proprietary architecture 3 Apple 4 Article Structure and Verifiability 5 "MESA CPU" 6 Smalltalk

[edit] Why Star was slow

Grumble. There are several unimportant inaccuracies in this article, but I bristled at the line "Finally, by today's standards, the software would be considered very slow, taxing the limited hardware of the era."

The software was so incredibly optimized, you wouldn't believe it. The problem was the lack of physical memory on this pathetic hardware platform. If memory serves, Star something like a half a megabyte of real memory. Programmers spent long hard hours "packaging" the software to minimize paging, but the poor thing was just constantly swapping itself to death. It was a nightmare.

If Star had had today's complement of 256 MB or more of RAM, it would have rocked. But 25 years ago, memory was expensive, and a Xerox executive made the bad decision that software optimizations could overcome the lack of memory. He was fatally wrong. - Rlw (Talk) 04:23, 16 September 2005 (UTC)

This is true. In fact, in the final version of this product the virtual machine had been ported so that it ran on top of Windows 95. It was MUCH faster than Microsoft Office, and much more powerful. Even with just 1.5MB of RAM the system ran well -- there was little swapping with that much memory. The initial CPU was about equivalent in performance to a Motorola 68000 processor. Since then we've seen at least 3000 times improvement in processor speed.

[edit] Proprietary architecture

I think I've read that you needed a special key from Xerox to write programs for the Star. Is this true? If so, I'd say that the proprietary architecture put it at a heavy disadvantage vs. the IBM PC. Perhaps this should be listed in the article as a possible reason for its failure.

Baccala@freesoft.org 05:26, 19 January 2006 (UTC)

No, you just had to have the Xerox Development Environment which included the Mesa Language (precursor to Modula and Java) and all the Interface Definitions. Starting in 1985 or 1986 this product was available for purchase -- and some major customers did write custom software for Xerox Star, well really Xerox Viewpoint -- the "open" version of Star that was released in 1985/6. However, like anyone at that time trying to learn to really program in Smalltalk, learning to program in Mesa could take as much as 6 months. It wasn't the language that was difficult, but vew people had been exposed to lightweight threads, memory mapped files, semaphors and object oriented programming in those days. And, it always takes time to learn a large number of libraries (Java, Smalltalk or Mesa) before you can do serious programming.

[edit] Apple

Nowhere is it motioned that Apple paid Xerox about 3 mil in stock options. I also wasn't aware that Xerox sued Apple. Was this part of the reason that Xerox lost the suit. Maybe someone more knowledgeable on this subject could clear this up.

[edit] Article Structure and Verifiability

Some parts of this article are way too informal. Sounds like a chat between buddies. I know wikipedia is for everyone etc. but this is something else.--MrBobla 00:35, 15 March 2007 (UTC)

I agree with MrBobla. I have no reason to think that this article violates NPOV (I evaluated a loan unit of the 8010/40 for several months in 1987) but there are too many unsourced statements to be compatible with the Verifiability and No original research rules. IMO this is important because the Star is popularly regarded as the prime mover behind the WYSIWYG and compound document initiatives by all major vendors, and this article will therefore be a first point of call for many students. It follows that the article should meet the highest standards demanded by Wikipedia. Until it does so, I recommend that it be tagged 'refimprove' as a warning to readers. Pointillist (talk) 23:32, 12 December 2007 (UTC)

User:Pzavon makes a fair point about 'I personally attended and heard this' being almost O.R. I was thinking "I'm sure this is true and somewhere I'll find confirmation, so this was just a placeholder". I am still searching for a verifiable source for the assertion that the IBM Displaywriter used the 8086 CPU. IBM UK said this at the launch and plenty of sites say so too, including Intel 8086 here on Wikipedia with no source cited (I've now marked that 'fact' to encourage others to find a source). BTW I've looked through much of the IBM Journal of Research & Development vol 25 issue 5 (at http://researchweb.watson.ibm.com/journal/rd/255//ibmrd2505B.pdf thru http://researchweb.watson.ibm.com/journal/rd/255//ibmrd2505ZI.pdf) but no luck so far. Pointillist (talk) 00:14, 22 December 2007 (UTC)

I am cerainly no expert in "the way things are done" in the Wikipedia, but I would have thought that one of the citation-needed templates would be an ideal citation place holder, and would not raise issues of original research, etc., for those not aware of the editor's placeholder intentions. Pzavon (talk) 04:14, 22 December 2007 (UTC)

[edit] "MESA CPU"

Prior to the mention of a "MESA CPU", Mesa was only referenced in the article as a programming language. Under Hardware description, though, I note that it does say "Its microprogrammed, bit-sliced CPU ran a virtual machine for the Mesa programming language" without naming this as a MESA CPU. Is this bit-sliced CPU indeed known as the "MESA CPU" ? --NapoliRoma (talk) 04:43, 11 March 2008 (UTC)

[edit] Smalltalk

It is unclear what Smalltalk has to do in this article. Search for the "smalltalk" character string and see for yourself. Thanks in advance for clarifying (or pruning).
--Jerome Potts (talk) 11:31, 20 March 2008 (UTC)