In article <tsmkhl$3djcl$1@dont-email.me>,
dxforth <dxforth@gmail.com> writes:
> On 17/02/2023 7:40 am, Gerard J. wrote:
>>
>> I just wanted to let everyone know that Mr. Emmanuel Roche is wrong. The book, on page 70, indeed shows the first page of the listing of Altair 4K BASIC. The credits are as follows (including the line numbers):
>
> Gates is famous for being the nerd who built a business empire and amassed
> a fortune. Being first to sell a working BASIC to Ed Roberts no doubt helped
> (allegedly Roberts was getting several calls a day from people claiming to
> have a BASIC 'nearly ready'). More interesting is what, if anything, was new
> about Gates' BASIC. If all the elements pre-existed, then perhaps not a lot.

I don't know about the earliest Gates/Microsoft BASIC, but by the time
of what came with my Osborne 1, it was an interpreter but tokenized,
where each reserved word was a byte with the high order bit on, and
some other things that internalized it enough to make interpretation
faster. Some googling may find an MBASIC program that can (with
minor adjustment for where in RAM the user program starts) list itself
without using LIST, by PEEKing through RAM from that starting point,
and printing a detokenized representation of itself.

Contrast that with CBASIC descended from BASIC-E (the latter by one of
the service academies, if memory serves) that had a separate BASIC to
byte code compiler program vs byte code interpreter that actually ran
it. Microsoft BASIC was definitely easier to use, although CBASIC may
have had some desirable features or performance advantage, which would
be way more than I remember.

So if Microsoft BASIC was easier (didn't need separate editor and two
more commands to compile and run), it might have had the advantage of
more people being willing to use it.