Mark Kent <mark.kent@xxxxxxxxxxx> writes:
> begin oe_protect.scr
> Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> espoused:
>> __/ [ Ray Ingles ] on Friday 24 March 2006 13:07 \__
>>> On 2006-03-24, Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> wrote:
>>>>| Up to 60% of the code in the new consumer version of Microsoft new
>>>>| Vista operating system is set to be rewritten as the Company
>>>>| "scrambles" to fix internal problems a Microsoft insider has
>>>>| confirmed to SHN.
>>>> This is still alleged to be an uncofirmed rumour.
>>> That's flat impossible for a codebase that large (wasn't XP around 40
>>> million lines?) if they have the slightest, dimmest hope of shipping
>>> *anything* this year or early next year. Especially given *their*
>>> Sure it's a rumor. They can't possibly do that, and so they'll end up
>>> cramming in as much new code as they can manage (which won't be anywhere
>>> near what they want, let alone need). They'll ship, regardless of the
>>> actual quality or readiness, and address the numerous problems with
>>> major patches, things that replace whole subsystems.
>> This turns out to be true; not just a rumour (spreading to media ATM). Think
>> of this as a plan, only a tiny subset of which will reach the Vista trunk.
>> It leave very little time for public testing. Rather than extend in the next
>> 5 years (Monad, WinFS, etc.), it seems like Microsoft will only play
>> catch-up by re-writing code. They already play catch-up with Apple and
>> Linux, feature-wise.
> Someone else noted that they're bringing the xbox360 programmers on
> board too. The problem here is that adding programmers is unlikely to
> solve their problems here... at least in a 1 year timeframe, which is, I
> presume, what they're planning. A good illustration of how long it
> takes to properly revise a large code-base would to look at how long it
> took the Mozilla team to move from Netscape 4/6 code to get to Firebird
> and Thunderbird and so on. There was some very good work done on
> Mozilla/Netscape 7, but it wasn't really the right target, whereas
> Firebird/Thunderbird have proven to be the right choice. I forget, but
> something like 3-4 years? And these were people doing it because they
> /wanted/ to, not just because they were being exhorted to by their
SW development has suffered from too many cooks spoils the broth for
as long as I've been involved with it. Bosses think that throwing more
people at it will speed things up. In most cases it slows things down
and leads to poor quality code.
Even today few SW managers understand the SW development
process. Sadly, many good SW engineers are promoted into management
positions as it is the only career progression path even though they
are not cut out to be managers. Everyone loses. They stop developing
good SW and then their bad management hinders those that are
developing the SW. I've seen this so many times it is truly pathetic.