On Windows’ technical quality, I found the Kuro5hin article on the Windows 2000 source instructive. (No code quoted, only comments.) This illustrates the Windows problem: smart people, doing their best, with something that was not designed.

(Relentless backward compatibility leads to its own problems, but others (notably Sun with Solaris) have less problems with this because their Unix actually has a design and modularity.)

I find this interesting as a particularly fascinating car crash. It’s the sort of thing that makes me think Microsoft will crash in a sudden and unexpected (except by those who were paying attention) manner – there’s really very little under the shell.

It’s the last twenty years, not the last two years. This has been how Windows has always been built: brilliant people doing things shoddily.

Actually, Windows was built as a bad copycat of another system.

Also see:

http://www.vanwensveen.nl/rants/microsoft/IhateMS.html
http://en.wikipedia.org/wiki/Gary_Kildall

Previously developers cycled out because they’d had enough options vest that they were rich; now they cycle out because the stock is moribund and Microsoft is increasingly clogged by bureaucracy (see Mini-Microsoft comments).

This is why Wine takes a pragmatic approach to win32, implementing only what’s needed for applications when a need is shown – rather than sitting down with MSDN and trying to implement everything in full from the book. Large areas of win32 are all but unused in real life and are pretty much superfluous.

This is what happens when your operating system is written by continuously swapped-out fresh graduates who are indeed smarter than anyone else but have no experience, the feature roadmap is dictated by the hairpin turns of marketing and your entire approach to gaining vendor lockin can be summed up as “layer violation:” you end up with a ton of stuff that no-one understands, no-one uses and no-one keeps working, assuming it ever did work (which it all too often didn’t – the documentation is frequently, ahh, more hopeful than accurate). This is why the Samba team are more expert in SMB than anyone left at Microsoft, for instance.