Miscellaneous Links 2 - The Vengeance

[hinode]

http://www.theaustralian.news.com.au/story/0,25197,24858964-12377,00.html

Terry Pratchett knighted.

[Fudozukushi]

http://www.theaustralian.news.com.au/story/0,25197,24858964-12377,00.html

Terry Pratchett knighted.

bitchin'

[Grefter]

http://www.rpgcodex.net/phpBB/viewtopic.php?t=26591

Mod released for Torment to run it in higher Res than 640x480.  This is pretty win.

[Shale]

http://www.zuneboards.com/forums/zune-news/38143-cause-zune-30-leapyear-problem-isolated.html

The code that caused the Zune bug. No, the programmers didn't ignore leap years entirely. Yes, they are still really, really stupid.

[Talaysen]

http://www.zuneboards.com/forums/zune-news/38143-cause-zune-30-leapyear-problem-isolated.html

The code that caused the Zune bug. No, the programmers didn't ignore leap years entirely. Yes, they are still really, really stupid.

Epic.  Yay for untested code~

[Twilkitri]

Because checking the last day of a leap year is the most obviously necessary test to run.

[Jo'ou Ranbu]

When you're two years away from one and you expect your product's lifetime to see at least one of them, yeah, I'd consider that important to run considering how badly they crashed.

[Twilkitri]

Unless you plan on checking every single date in the next X time, you only check the ones that appear relevant and potentially a random scatter. And the only relevant dates as far as leap years are concerned are February 29 and March 1.

This should have been caught in review, not testing.

[Talaysen]

Unless you plan on checking every single date in the next X time, you only check the ones that appear relevant and potentially a random scatter. And the only relevant dates as far as leap years are concerned are February 29 and March 1.

This should have been caught in review, not testing.

And December 31.  Especially if you look at that function.

[Twilkitri]

A logic error in that particular function does not make December 31st in and of itself any more testworthy specifically in leap years.


On review, I guess that ensuring December 31st doesn't roll over to the next year could be potentially beneficial to have tested. IF the date format allows for that to happen in the first place, which we can't see in the code provided. But that doesn't change that the error should have been caught in review, and that you have no way of knowing whether they did any testing or not.

[Shale]

On review, I guess that ensuring December 31st doesn't roll over to the next year could be potentially beneficial to have tested. IF the date format allows for that to happen in the first place, which we can't see in the code provided.

It does allow for that - the post clearly states that the Zune calendar works by calculating the number of seconds since a baseline day (Jan. 1, 1980), and then chopping the seconds into minutes, hours, days, months and years. Make a mistake in the functions that calculate the varying number of days in a month or year, and you'd get a calendar that rolls over at the wrong time. The other side of that would be that you'd think an error in leap year handling would cause the dates to be wrong no matter when you checked - seeing that the function correctly accounted for the leap years in previous years would seem to clear it.

Review still should have caught the mistake, though.

[Xeroma]

And with that, *closes thread for new one*

[metroid composite]

Eh, even though this topic is closed, may as well keep the zune reply in this one as opposed to being a reply-to-nothing in the new topic.

Unless you plan on checking every single date in the next X time, you only check the ones that appear relevant and potentially a random scatter. And the only relevant dates as far as leap years are concerned are February 29 and March 1.

This should have been caught in review, not testing.

Umm...I think you underestimate the budget of QA testing.  It's fairly standard to test every relevant date within a certain range.  Assuming it takes a tester 1 minute to change dates, checking every date for the next two years is approximately 12 hours of work (which is to say, one day's work in the life of a tester, so probably costs Microsoft about $100).  Seriously, we had a tester collect 2160 collectible spider tokens in Spiderman: Web of Shadows--testing 750 dates is comparatively tame.

What I do wonder, however, is that from the description it sounds like the bug kicks in when you boot up the Zune; would the tester have missed the bug if he didn't turn the Zune on and off after every date switch?

[Shale]

That varied, for some reason - some of them hit an error when the date changed, rebooted immediately and got caught in the infinite loop, others continued as if nothing was wrong, but would hit the loop if turned off and back on. Mine was off overnight, so I don't know what would have happened to it.

[Grefter]

http://www.youtube.com/watch?v=v4kswfVgoLU&feature=rec-r2

Insert generic rage filled diatribe here and assume you all posted something retarded in response for the next 15 posts until Jim came in and was right.
(Monkey Magic theme song played on Piano, Cello and 2 Violins (too lazy to check if it is Viola in there or not), so it is cool)

Edit - My hopes are dashed, I was dreaming of them doing an Axel F cover.

[Xeroma]

Can a mod like OK come in here and actually close the thread properly?

[Grefter]

Nevar!

[AAA]

Uh I thought you could close your own threads.

[Halbarad]

Closed.