[Gllug] Controversial Joel Spolsky article

[Christopher Hunter]

On Sunday 21 Dec 2003 11:05 pm, Christopher Currie wrote:

> Hmmm. The original 8K and 12K Basic interpreters, which Gates & Allen wrote
> themselves, were pretty reliable. Once they'd expanded, of course, by about
> 1980 they had the defects you describe.

Gates actually didn't ever write any code - he just claimed he did!  Their 
"original" basic interpreter actually had several bugs or varying 
significance, so it CAN be safely said that M$ NEVER released properly 
working code!

> The case I was thinking about was in about 1983 or 1984. We had DEC Rainbow
> systems (CP/M 80 and 86 or MS-DOS 1!) and wanted the new Microsoft Basic
> compiler (v.5.36 I think), which was advertised as handling 16-bit strings,
> for simple text processing work. Bought it through DEC.
>
> When I wrote and tested a program for our job, it kept crashing at
> different points with a corrupt string-space error, making me think I'd a
> hardware problem.
>
>  I wrote a simple module to test the correct handling of 16K strings and to
> demonstrate the fault, and then complained to DEC, sending' em a copy of my
> module. They at first refused to do anything about it, but eventually
> admitted the fault; but they took 9 months to get M$ to admit the fault.
>
> By that time M$ had issued a new version of the compiler, still with the
> bug in it. They told us (via DEC) that we would get no refund,
> compensation, or discount on an upgrade; the next version of the compiler,
> which would come out a few months later, was going to be released with the
> bug still in it, but they'd fix the one after that and we could buy it at
> the full price.....
>
> Meanwhile I'd rewritten my program with the usual tiresome kludges to get
> round the 255-byte string limit, so that it was messy, took much more code,
> and ran more slowly. But we were still using it in 1999.

Doesn't that demonstrate the attitude of M$?  Nothing much has changed!  The 
original (and trivial) abort / retry / fail bug in DOS has endured through 
over twenty releases!

An ex-microserf of my aquaintance was privy to some high-level corporate 
meetings at M$, and described the contempt with which Gates and his board 
treat their clients.  They actually don't care that their "product" is 
fundamentally insecure, badly broken and mainly made of code stolen (and 
broken) from outside M$ (like "their" TCP/IP stack from BSD).  They're still 
making huge amounts of money and like to refer to the "GQ" (gullibility 
quotient) of their customers!

Chris

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug