MS-DOS Not Stolen, New Forensic Analysis Concludes 286
Posted
by
timothy
from the sorry-to-disappoint-you dept.
from the sorry-to-disappoint-you dept.
theodp writes "Challenging earlier assertions that Bill Gates got the rewards due Gary Kildall, a forensic analysis conducted for the latest issue of IEEE Spectrum concludes that the landmark MS-DOS operating system which Bill Gates and Microsoft licensed to IBM was an original piece of work, not stolen goods. Using his company's CodeSuite forensic software, Bob Zeidman said he found no evidence that QDOS or MS-DOS was copied from or was a derivative of Gary Kildall's CP/M. So, what do you think of Microsoft expert witness (pdf) Zeidman's "if-the-codebase-doesn't-fit-you-must-acquit" arguments?"
"The smearing of a computer legend" (Score:5, Informative)
Literally in the process of reading a dismissal of that same analysis. [theregister.co.uk] See what you think...
original piece of work: but not by MS (Score:4, Informative)
Always thought the issue was that MS did not have a license to re-license (e.g. to IBM) the product which was created by 'Seattle Computer Products'.
Re:What? (Score:4, Informative)
It means:
Challenging earlier assertions that Bill Gates got the rewards [that were] due [for/to] Gary Kildall
And yes, it's standard English.
Re:"The smearing of a computer legend" (Score:4, Informative)
That was for backward compatibility with existing CP/M code. By DOS 2.0 file handles replaced File Control Blocks.
Re:meh (Score:5, Informative)
I don't think so.
Microsoft did purchase Qdos, and there are direct correlations between qdos and msdos. What the article asserted was that there is no cp/m in qdos, nor is there any cp/m in msdos. There certainly is plenty of qdos in msdos.
alike and different (Score:5, Informative)
QDOS was actually quite similar to CP/M in it's structure, and CP/M86 was different in that it actually made use of the improvements offered in the 8086 processor. QDOS was written as if an 8080 to 8086 translator had been used to code it. However MS-DOS quickly moved away from this. What Microsoft sold was much polished over the original QDOS and CP/M OS's. They quickly improved the disk structure, FAT12 and FAT16 are different enough from the original CP/M disk structure. What they all STILL have in common is the use of the 0XE5 IBM uninitialized data marker in the FAT to show available space. This was a quick and dirty hack that allowed a freshly formated diskette to be used without having to initialize a directory structure on it.
Re:Alternatively... (Score:5, Informative)
This.
CP/M is a very simple beast. It's laughable to think that anybody would go to the effort of disassembling it to find out how it worked then rewriting it function-for-function in 8086 assembly code. changing the file system as you go.
It would be much less work to just read the CP/M docs then write your own little OS using the ideas gleaned. I doubt he even did that. There was no magic in CP/M even way back then and MS-DOS isn't all that similar to it.
Re:"The smearing of a computer legend" (Score:5, Informative)
The first 26 system calls of MS-DOS 1.0 are identical to the first 26 system calls of CP/M."
So? Windows NT has a POSIX interface in it somewhere. It's done to try and tempt people to port their POSIX code to windows.
Making CP/M code easy to port to MS-DOS would have been a good idea and those functions would have been needed anyway so arranging them in the same order is no real extra effort.
Re:original piece of work: but not by MS (Score:2, Informative)
Always thought the issue was that MS did not have a license to re-license (e.g. to IBM) the product which was created by 'Seattle Computer Products'.
MS had bought all rights to it from SCP before the launch of the IBM PC. http://en.wikipedia.org/wiki/86-DOS#Creation_of_PC.C2.A0DOS [wikipedia.org] . This wasn't disputed by SCP, but they claim they wouldn't have sold it as cheap if they had known about the deal with IBM and pending launch of the IBM PC..
Wait, what? (Score:5, Informative)
Where did this idea ever come from? "Everybody knows" that Gates bought QDOS from Kildall and nobody ever claimed that QDOS was a "copy of CP/M," not even Kildall himself. What was in dispute was whether Kildall was literally out to lunch or flying, or whatever, brushing off the meeting and selling a license for what turned out to be a pittance. That's the legend anyway, but he's not here on this planet anymore to defend himself.
FFS. Want to know where DOS came from? Just read Tim Patterson's blog. http://dosmandrivel.blogspot.com/ [blogspot.com]
He discusses the design differences between the two and why he did what he did.
And if you're really curious and need to feed the inner nerd, go have a look at the CP/M source code.
http://www.cpm.z80.de/source.html [z80.de]
--
BMO
Re:Alternatively... (Score:5, Informative)
More likely CodeSuite is right
More likely, people haven't understood the original dispute. Did QDOS steal lines of code from CP/M? Most likely not, but nobody ever claimed it did. Was it a rip-off of CP/M? Absolutely. QDOS implemented calls identically to CP/M with the specific aim of being as close to CP/M as possible. In other words, as Patterson him self said, he read through Kildalls manual and tried to create something that functioned identically.
As you point out however, he did a much better job on the FS, which is both to be commended, and also should be added on the "it was not a rip-off" side. DOS was an interrupt handler, and not much more though. As an interrupt handler it clearly "ripped off" CP/M to the point of being almost identical. However, not by stealing code. No stealing of code would have been needed (as you say) and that has never been asserted either. Not by the parties involved.
Re:Alternatively... (Score:5, Informative)
I'm about 90% sure you're wrong on this. CP/M's file system doesn't really resemble FAT at a low level. FAT actually originated as the file system implemented by Microsoft's early stand-alone BASIC implementations.
CP/M's file system was extremely crude, being comprised largely of a single table containing the directory, with the rest of the disk containing the data. Each directory entry contained pointers to up to 16 blocks, if a file contained more than that it had to have multiple directory entries.
FAT has multiple tablers, and FAT's system is, ultimately, based upon chains of clusters, and the directories simply point at the starting cluster, with the information about the groups of clusters themselves stored in a separate table (the actual FAT.)
They're both hideous, but they're not remotely similar.
I never said anything remotely similar.
What I said, which I believe is true, is that CP/M had to read the whole directory to find out what parts of the disk were empty. Many of the optimizations that occurred between 1.3 and 2.x were to address this specific issue.
Re:that was my understanding (Score:4, Informative)
Who is this "everyone" that is "surprised" that you only license Windows? Even with FOSS, you only have a license - you do not "own" the code. There can be only one owner of a thing, and in the case of Windows, that owner is Microsoft.
Re:that was my understanding (Score:4, Informative)
The people of the United States, as embodied in the federal government. One owner. You, as an individual, do not own the Washington Monument or any part of it.
Fact checking. (Score:5, Informative)
The fact remains that he did die conveniently in a plane crash just after failing to come to terms with MS.
Not true.
On July 8, 1994, Kildall fell at a Monterey, California, biker bar and hit his head. The exact circumstances of the injury remain unclear; however, he had suffered problems with alcoholism in his later years. Various sources have claimed he fell from a chair, fell down steps, or was assaulted because he walked in to the Franklin Street Bar & Grill wearing Harley-Davidson leathers. He checked in and out of the hospital twice, and died three days later at the Community Hospital of Monterey Peninsula. The coroner's report identified the cause of death as blunt force trauma to the head. There was also evidence that he had experienced a heart attack, but an autopsy did not conclusively determine the cause of death.
Gary Kildall [wikipedia.org]
Concurrent CP/M 3.1 and later, and single-user CP/M-86 with BDOS 3.3 and later (including DOS Plus), allow CP/M programs to access DOS-formatted discs via conventional BDOS calls, emulating (as far as possible) the behaviour of a normal CP/M filesystem. The behaviour is probably a good starting point for anyone writing a CP/M emulator which uses a hierarchical or non-CP/M filesystem.
The FAT filesystem in 16-bit CP/M-86 [demon.co.uk]
To me this says that the original or "normal" CP/M file system was not FAT.