history is hard (was: Microsoft open sources GWBASIC)

[Fred Cisin]

On Sun, 24 May 2020, Jecel Assumpcao Jr via cctalk wrote:
> I had heard that Microsoft had licensed Xenix before the IBM thing.

I hadn't known that.

> Bill thought he had a gentleman's agreement with Gary to not intrude in 
> each other's turf and then DR came out with CBASIC. Furious, Bill got 
> into operating systems in retaliation.

Gordon Eubanks (CBASIC) and gary were pretty tight.  I just looked up and 
found out that Gary was his thesis advisor in 1976, when he wrote CBASIC. 
It seems like it was inevitable that Gary would end up marketing it.

I played with it a little in mid 1980s; my recollection was that it was 
expensive, and impressive, but not especially suited for any of my 
projects.

> When IBM came to Microsoft for an OS they had specs for a machine that 
> was in no way up to running Xenix. So it is just simpler to tell the 
> story as "Microsoft didn't have an operating system". Adding DOS 
> complicated things for Microsoft so they planned to evolve the two 
> systems towards each other until there was a single one. The January 
> 1982 Byte says instead that there would be 3 systems: Xenix at the high 
> end, DOS at the low end and a hybrid in the middle. MS-DOS 2 was 
> essentially this hybrid (so most system calls have two versions: a CP/M 
> one and a Unix one).

I played briefly with Xenix on an XT (or MAYBE an AT) on a 15MB? drive 
partition.   MS-DOS was a better match for that hardware.

OS/2 (Gordon Letwin at Microsoft) was a substantial step up for MS-DOS.
Once they added "Windows For Os/2"/"Presentation Manager", . . .
BUT, then NT was not a direct transition from OS/2.
And, around 1986? IBM started pushing OS/2 with PS/2 (had they bought OS/2 
from Microsoft by then?)


MS-DOS was based heavily on CP/M.  Most university programming 
graduates were into unix.  When Microsoft or Apple were recruiting, 
most of the best pickings were C programmers on unix.  MS-DOS 2.00 was 
definitely moving towards unix, in terms of the sub directories, and file 
handle based API.
I didn't know that there were plans for three levels, just that they were 
moving MS-DOS towards being unix-like.  Keeping the CP/M style API kept 
most software compatible.  FCBs made parsing filenames in the command line 
convenient!


> Hmmm... here it says that though CBASIC was developed in 1976 it only
> became a DR product in 1981, which is too late for the story I told
> above to make sense:
> https://en.wikipedia.org/wiki/CBASIC

Well Gary and Gordon were close in 1976, so you could use any date in that 
1976-1981 range.

> I have just watched a talk with Gary where he introduced DR Logo. It
> wasn't much cheaper than that. And at the end somebody asked about the
> soon to be introduced C compiler and the answer was that it would cost
> $600.
> https://www.youtube.com/watch?v=l4P6MDuk3Zk

Yeah.  some software was free, or almost, and some was very expensive. 
$240 for CP/M-86 was not out of line for the time, other than the fact 
that it was going head to head with an already established $40 (later 
$60?) product.  A few years later, DR dropped the price of CP/M-86 down to 
$60.  TOO LATE.

>> BTW, once Microsoft started work, IBM insisted on upgraded security and
>> locks.  For a while, it was referred to as "Project Commodore" as a red
>> herring for any leaks.
>
> They installed barbed wire in the air ducts going into the room with the
> prototype.

I hadn't known that, but it fits with an IBM PHYSICAL SECURITY attitude.
(and totally out of character for west coast software culture!)

--
Grumpy Ol' Fred     		cisin at xenosoft.com