Hey Scott! Too Dumb to Date Sort? or "When is 27 > 28"?

Its been a day down in the land of gnarly bugs.  No that isn't the land down under (unfortunately).  Instead its been a day of dealing with a wee bit of miscalcuations on my part.  You may have noticed some rather "interesting" date sorting recently and a situation where an article says published March 27th but is sorted above March 28th in a descending order date sort.  Clearly something is less than correct. 

Dates are fun things in feedster -- I have so many of them!  There's the date published (pubdate) and the date indexed (epochadded).  The general case where the date sorts are wrong is tied to the case when I got from the pubdate field of the RSS item a value like this: 2003-03-27T17:27:00-08:00.  Now that's a Movable Type'ism and I'm not going to say if its right or if its wrong.  Personally I would have preferred something that the standard GNU time and date libraries don't barf on, returning '-1' but that's just me.  And I'm pragmatic -- whether or not this is right or wrong, its in wide spread use so Feedster should support it.  Period the end.  I know that.  So this led me into the land of parsing dates, looking up time zones, thinking about offsets and such. 

Well I've emerged from the Land of Gnarly Bugs and I feel **much** better for it.  I'm now properly handling Movable Type dates and things are better.  Example of a date sorted search.

Now I won't lie to you and say that the results are perfect (yet).  A number of database entries are lacking a well formed pubdate field and may, at times, default to the date indexed rather than the pubdate.  Hopefully that should be minimal.  Additionally I should really display a normalized date for each result rather than the hodge podge of different date formats that each blog uses.  That'll come (but it will be a bit).

And the award for "Most Helpful Guy on a Thursday Night All the Way from Holland" goes to Rob Kaper of The Amazing - US who not only made me aware of how large an issue this was but also pitched in on getting it solved as well as tolerating an american's natural stupidity when it comes to time zones beyond EST, CST, PST.

Around all weekend and grinding hard ...

4:56:16 PM      Google It!   comment []    IM Me About This   

Google on Indexing Your Hard Drive

Interesting tidbit here from the interview with Sergey Brin:

Will Google provide a desktop utility -- e.g. index/search my drive?

He says no, it's not a core competency and no one has ever made it work that well.

Interesting.  So the kings of search don't think it works that well.  And why do we think then that Microsoft's Longhorn with embedded searching will work that well?

11:21:39 AM      Google It!   comment []    IM Me About This