Microsoft has always thought Betas as just another stage in bug hunt: the product, now stamped "good" after the rounds of internal tests, then made FC (Feature Complete), goes out of factory, to meet the users who want help finding the most possible of the probable remaining bugs and misconceptions. Hence the Beta *MUST* remain loaded with debugging/reporting code (no automatic SDK can totally remove the need for this).
This bug hunt still may come back with some deep changes to make in the code; so it's not worth (in MS mind at least) to invest time and effort in optimizing that code at that stage; so, that Beta *MUST* remain slow during all the time of its existence.
When the product is considered debugged enough, then only will MS decide to stop bug hunt and start cleaning the code (from all the debugging/reporting stuff), then optimizing it.
When that optimization is advanced enough that the product could actually go to publication, they check it is still sufficiently bug-free, and if so, it gets the RC stamp (Release Candidate) - that is, it gets published as another Beta, but this one is just for final tests, during which it is ready at any moment, if test results are sufficient, to be called the final product, and sent to mass CD burning.
This way of thinking, I don't share it (I produced by myself big applications that, working the way big software companies work, would never be possible to write with the same level of reliability, productivity, friendliness), but I respect it (commercial and technical advantages/drawbacks can be weighted). It implies a strict discipline in the team: in a large team, when the debugging/reporting code has been removed, developers must resist from any temptation to make any significant change - or the product could get hit with a new cycle of internal debugging, external beta, RC. That is, even the optimization has to remain limited - or it would need too deep code change, implying too high risks of a new cycle.
In addition on the marketing side, MS needs its product has some commercial momentum at the right moment when it actually ships. Hence it's better to keep its reputation artificially low during the months before the period of time when it must be ramped up. The current "comments" about "Vista being a hog" may very well be part of such a scheme (they could be not either - I have no proof on any side, I remain open, and above all this is not my main interest point).
Of course I am no more MS insider than others, so this scheme (as others') is just a guess from me, built after observing (and remembering!) what happened with W3.11WG, W95, W98, W98SE, W2K, WXP (I didn't use 3.0, NT or ME).
Paris, Wed 3 May 2006 17:23:10 +0200
|>> Re: FC Beta must remain slow - RC must be faster
I'd have said it differently, but not better.
Beta code has debug latches built in that will detriment overall system speed.
I must point out that the general concensis is that Microsoft has determined that "the world is our beta site".
In it's (Microsoft's) defense, I also have to point out that they are attempting to provide broad based services to those that have no or little computer based skills. They an Apple are attempting to make a specialized device into a general purpose consumer commodity....thus the rub.
Ad Astra Per Aspera
(A rough road leads to the Stars)
We all know what we know, and everyone else knows we are wrong.
System Specifications in BIO