Re: PostgreSQL - 'SKYLINE OF' clause added!

<p><font size="2">Yep - we're paying attention Josh!<br /><br /> I like the category being explored with skyline, I'm
notsure yet how it fits with existing 'soft data' models and applications that use them.<br /><br /> If SKYLINE is
interestingto app developers, maybe we should consider it for Bizgres?<br /><br /> - Luke<br /><br /> Msg is shrt cuz m
onma treo<br /><br />  -----Original Message-----<br /> From:   Gavin Sherry [<a
href="mailto:swm@alcove.com.au">mailto:swm@alcove.com.au</a>]<br/> Sent:   Wednesday, March 07, 2007 05:44 PM Eastern
StandardTime<br /> To:     Josh Berkus<br /> Cc:     Tom Lane; pgsql-hackers@postgresql.org; Alvaro Herrera; Chris
Browne<br/> Subject:        Re: [HACKERS] PostgreSQL - 'SKYLINE OF' clause added!<br /><br /> On Wed, 7 Mar 2007, Josh
Berkuswrote:<br /><br /> > "Approximate queries" is something with DSS users *want*.  Jim Grey addressed<br /> >
thisin his ACM editiorial on the databases of the future.  It's something<br /> > that *I* want, and if the
Greenplumpeople aren't speaking up here, it's<br /> > because they're not paying atttention.<br /> ><br /> >
Now,I don't know if this Skyline patch is our answer for approximate queries.<br /> > Maybe I should pester Meredith
aboutgetting QBE free of its IP issues; it<br /> > certainly looked more flexible than Skyline.  In either case, the
code<br/> > probably needs a complete refactor.<br /><br /> What people want from "approximate queries" is different
tothis: the<br /> desire is usually to balance run time with level of accuracy/quality (some<br /> times the desire is
tohave accurate results as well as similar results).<br /> Neither skyline or QBE are about this. The only thing in the
specwhich<br /> addresses this is 'tablesample'.<br /><br /> Thanks,<br /><br /> Gavin<br /><br />
---------------------------(endof broadcast)---------------------------<br /> TIP 9: In versions below 8.0, the planner
willignore your desire to<br />        choose an index scan if your joining column's datatypes do not<br />       
match<br/><br /></font>