Intel's lousy software III – MS says it has enough of its own
Sunk in VXD hell, Microsoft felt it just couldn't deal with any more crap...
When the claims that Microsoft had strong-armed Intel into dumping NSP first arose, Microsoft retorted that it had opposed NSP because of the low quality of Intel software. Bill Gates took the lead on this (see Intel writes lousy software, says Gates). The line is being repeated now by Gates' lieutenant, Paul Maritz, who says: "Unfortunately, because the PC software produced by Intel is mainly done to 'jump start' the demand for new hardware, it often falls below the high quality standards of Microsoft software. Testing -- a vitally important part of the software development process -- is often insufficient. Moreover, Intel often devotes insufficient attention to providing compatibility between its software products and other software products. Providing such compatibility is a major focus of Microsoft's development efforts." Silly old Intel, thinking it can do amazing things with a couple of dodgy little hacks and no back-up. But as we pointed out a while back (see Lousy Software II), Intel itself claims title to large swathes of the software innovations in Windows 98. And if you check out the email traffic between Gates and Maritz at the time of the NSP spat, you get a slightly different picture. There were issues as regards Intel software development, the most specific one being that Microsoft -- perfectly rationally and reasonably -- thought that Intel's multi-unit and unco-ordinated approach to software was difficult to work with, and difficult to make work. Pot and kettle though -- Microsoft at the time had numerous different bits all working independently on Web-enabling diverse Microsoft products, and one of Maritz's headaches was getting them all to point in the one direction. But what about NSP support? Microsoft today says it wasn't supporting NSP because Intel's software was bad, and because it didn't support Windows 95, which was due out in a few months. But from the email traffic it's clear that Intel was lobbying Microsoft to incorporate NSP support in 95. Maritz, however (May 1995), is having trouble with dealing with this: "I fear that: we have a baroque enough mess already in our MM plumbing without having to stitch iSPOX in." Microsoft's problems with its plumbing would unfortunately appear to be ones of its own making. Says Gates as part of the same exchange: "Unfortunately our wide open VXD architecture is causing us immense problems and is just going to get worse in general." So it would appear Microsoft Windows development was at the time sinking into VXD hell. It's puzzling stuff, particularly if you check out Maritz's explanation in his deposition of why Microsoft succeeded while IBM and OS/2 failed. He points out that Microsoft took a twin-track approach to operating systems, going for Windows and NT at the same time, and that if IBM had taken different decisions at the time, the outcome would have been different. The reality of course was that VXDs were a dodgy hack used to implement 32-bit in Windows in a way that was massively inconvenient and destabilising to OS/2 (that's not to say IBM wouldn't have screwed-up anyway, of course). But three years on, it would appear that "our wide open VXD architecture" is coming home to roost. ® Complete Register trial coverage
Sponsored: RAID: End of an era?