Slashdot Mirror
Bug Pushes Vista Out to November 8th
IntelliAdmin writes "Microsoft originally targeted October 25th for Vista's release to manufacturing, but a last-minute bug that 'took most of the Vista team by surprise' has caused an unexpected delay, said Ethan Allen, a quality assurance lead at a Seattle high-tech company that tests its products for Vista. Allen said the Vista team discovered the bug, which 'would totally crash the system, requiring a complete reinstall'. Vista now has a new RTM date of November 8th" A reader wrote in to point out this story originated with Paul Thurrott.
I'm sure all kinds of jokes about MS bug history will come up, but at least they caught it before it was officially released. Better a 2 week delay to fix the problem than them saying they will worry about it later in an update.
That said, this sounds like a fairly major bug to catch this late in the game.
A crash is one thing, but a re-install to fix it? I have my doubts, but if anyone can pull it off, it's Microsoft!
Not that hard to imagine, really. A filesystem driver bug that blows away critical tables in the filesystem could put you out of commission pretty quick. (I have no idea what the bug is but filesystem corruption is the most likely thing I can think of.)
That's a pretty nasty bug! That's almost as bad as a bug that could start your computer on fire, or punch you in the face. I don't understand, they would pump this software out with the bug had someone not stumbled across it? Ew.
Don't blame their testing practices, blame their development practices. Bugs which occur randomly 0.001% of the time are extremely hard to catch by testers. The key is to use coding practices which make it even less likely that you will CODE such a bug. Memory corruption and subtle synchronization errors are the most common sources of bugs that can be extremely hard to reproduce. Rather than testing for such bugs the strategy should be to use coding practices that guarantee no memory corruption and no synchronization problems. In reality, people get lazy and code bugs that occur once in a decade, and when they are caught by testers at the last second guess who gets blamed?There's no telling how many OTHER bugs are still there, waiting to totally crash your system and force you to reinstall. We can make guesses (200? 500? 1000?) but nobody really knows.
$nice = $webHosting + $domainNames + $sslCerts
Why are you surprised at this?? Do you know how many other things seem to require a complete re-install?
Every time I hear someone who supposedly knows a lot about Windows tell me to reboot the machine all of the time, or say "dunno, maybe you should re-install", I just want to choke someone. But the first Microsoft patch (reboot) is frequently followed up with the second (reinstall).
I've seen numerous machines that have become so unuseable as to require the scorched-earth approach.
Cheers
Lost at C:>. Found at C.
How can the majority of their customers use it if it hasn't even been released yet?
Rhymes that keep their secrets will unfold behind the clouds.There upon the rainbow is the answer to a neverending story
Once again, some Slashdot users prove that their hatred towards Microsoft surpasses objectivity. The article does not say how this bug occurs, how often or even why, so for all we know, this could be a very uncommon bug. It's just a good thing if the quality assurance team spots a bug and eliminates it, right? Why on earth should we flame them for that? As if the development of Linux was flawless?
I for one say, let's judge the final product before we smack Microsoft for something that's not yet released to the public.
Full Tilt
Oh yeah, that's the OS I want to base my Internet and personal business on. A total meltdown bug that takes most of the huge OS team by surprise on the day it's supposed to be manufactured ("in stone"), after all the testing is supposed to be complete. But it doesn't surprise everyone, so it's been known to some on the team - but slipped past testing anyway. Which causes a delay of only two weeks, despite the testing necessary to be sure this bug 1: is gone; 2: doesn't break anything else when fixed; and 3: doesn't have others like it waiting to "surprise most people".
What kind of $MULTIBILLION corporation, whose steady stream of "upgraded" products are essential to global business and billions of personal lives, runs this way?
Microsoft. When monopoly is all you need.
--
make install -not war
This bug 'took most of the Vista team by surprise'. Well, gee-whillikers, I would hope so. If most of them _weren't_ surprised, I'd be even more dismayed.
Sigh.
Now, Steve Bink at bink.nu is a great guy and a friend, and I know he had no idea that these guys were just ripping me off. But that's the point of this: If you separate a story enough from its true source, it's becomes kind of unclear what the truth is.
Welcome to my life.
Poor Paul Thurrott! Such a hard life you lead. "I wrote about this first, I wrote about this first! digitimes didn't credit me! IDG credited digitimes, not me! I wrote about this first! bink linked to the IDG story, what about me!"
Paul Thurrott may be an important figure in the coverage of Microsoft Product or something, but I hardly think he's the only person with "sources" who get tipped off when these things happen. Maybe, just maybe, digitimes has sources too, and they found out about the setback from some place other than Paul Thurrott's site(s). Paul needs to get over himself, he's not the sole source of Microsoft news.
I concur, I'm so glad Apple did the same thing with Mac OSX.
Hmmm... when I used to repair windows machines it was probably my #1 task. Many problems probably could have been fixed but customers didn't want to pay $65/hour to have me remove and reinstall drivers and apps, hack at the regisrty, or whatever else. Actually I rememebr a lot of technet articles that atcually said that reinstallation was the recommended solution. (this was pre XP days, I haven't been in that business since XP was released)
Just because the bug is severe doesn't mean it's easy to reproduce. It may happen in one very specific set of circumstances. You could test for 100 years, and if you never hit that one, specific case, you'd never see the bug.
The number of possible scenarios in something as complex as an OS is *staggering*, you just can't cover every last case with any reasonable amount of time and manpower. So, you design tests to cover sensitive areas and likely trouble spots, you take as large a sampling of other cases as possible, and you accept a certain amount of risk. Sometimes, someone gets lucky and stumbles across a showstopper two days before you release. Better to have found it in-house than to have a customer report it.
I don't remember the source, but I read once that the probability of introducing a new bug in the process of fixing an old bug was somewhere between 20 and 50 percent. My experience as a software engineer would push me to the lower end of that estimate.
So, what did they do on October 13? Add a huge new feature? Or just fix a bug that was bad enough that they felt they had to fix it before they shipped, and make a mistake in the fixing process? My bet would be on the latter.
I mean, I hate Microsoft as much as the next Slashdotter, but I doubt they were dumb enough to be adding anything but bug fixes that late in the game...
Side note: The two scariest releases I have ever been part of took place on Friday the 13th and Halloween, respectively. And they were not scary because of the date, they were scary because we were nervous about whether the code was solid. Both went off without a hitch. So don't blame the date on this one, either.
WTF are you on about? Do you have any experience in computer science at all? Because you're speaking utter rot.
Automated tests are better.
Automated tests can be run at night, when no one's around. They can be run constantly, without driving someone insane.
Automated tests are reproducible. Try following someone's 'Uh, I clicked here, then opened this, then I think I cancelled that program, then...' instructions a few times. Then tell me automated tests aren't preferable.
Can't keep up with all the tests to run? Buy a new computer. Your scheme would have a new person hired every time someone's maxed out. (Or, alternately, dumping old tests.)
Automated tests cover regressions. Found a bug? Write a test for it. Then if it pops up again (which they always do), you catch it early.
Automated tests can be run by anyone, if done properly.
Automated tests are predictable. They do, in fact, cover the same code each time. This is an asset, not a liability. You know exactly what you've tested, and what you haven't. You can write _more tests_ to cover the other stuff. You'd rather someone happen to click a little different on the last build, and miss a regression?
Manual testing is required for GUIs to some extent, and to winkle out usability issues.
To suggest MS is dumb because they tried to make their testing rigorous, predictable and regular is utterly absurd.
Finally, a reasonable post.
The vast majority of the posts on this subject leads me to believe that the vast majority of slashodtters don't have the first clue about the development and testing of a large project.
Oh, and let's not forget that a few months ago an Ubunto update deleted the entire home directory of users. That's as major as this Vista bug, and was readily producible (unlike this Vista bug), yet it slipped through.
-- "I never gave these stories much credence." - HAL 9000
Your link doesn't seem to list any bugs requiring re installation.
Every OS has bugs. But there are bugs, and there are BUGS.
"There is more worth loving than we have strength to love." - Brian Jay Stanley