Обсуждение: Old-style OR indexscan slated for destruction
I am about to rip out the code that supports multiple indexscans for OR conditions inside a single IndexScan plan node. As best I can tell, the new-style bitmap-OR code is as fast or faster than the old way even in fully cached test cases (ie, with no allowance for improved efficiency of disk access). So there's no percentage in maintaining support for the old way, and getting rid of it will allow simplification of code and data structures in the planner. regards, tom lane
On Sun, 24 Apr 2005, Tom Lane wrote: > I am about to rip out the code that supports multiple indexscans for OR > conditions inside a single IndexScan plan node. As best I can tell, > the new-style bitmap-OR code is as fast or faster than the old way > even in fully cached test cases (ie, with no allowance for improved > efficiency of disk access). So there's no percentage in maintaining > support for the old way, and getting rid of it will allow simplification > of code and data structures in the planner. very good, Tom ! > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 6: Have you searched our list archives? > > http://archives.postgresql.org > Regards, Oleg _____________________________________________________________ Oleg Bartunov, sci.researcher, hostmaster of AstroNet, Sternberg Astronomical Institute, Moscow University (Russia) Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/ phone: +007(095)939-16-83, +007(095)939-23-83
>> I am about to rip out the code that supports multiple indexscans for OR >> conditions inside a single IndexScan plan node. As best I can tell, >> the new-style bitmap-OR code is as fast or faster than the old way >> even in fully cached test cases (ie, with no allowance for improved >> efficiency of disk access). So there's no percentage in maintaining >> support for the old way, and getting rid of it will allow simplification >> of code and data structures in the planner. For all index types? Even lossy ones? Chris
Christopher Kings-Lynne <chriskl@familyhealth.com.au> writes: > I am about to rip out the code that supports multiple indexscans for OR > conditions inside a single IndexScan plan node. As best I can tell, > the new-style bitmap-OR code is as fast or faster than the old way > even in fully cached test cases (ie, with no allowance for improved > efficiency of disk access). So there's no percentage in maintaining > support for the old way, and getting rid of it will allow simplification > of code and data structures in the planner. > For all index types? Even lossy ones? Can't see that a lossy index would make any difference ... regards, tom lane
>>For all index types? Even lossy ones? > > Can't see that a lossy index would make any difference ... OK, was just checking :)