easier COUNT(*)

[jek]

func.count() with no arguments should emit COUNT(*)

[Mike Bayer]

well it looks like my ancient wisdom on this is wrong, though I can't find the original source:

http://www.thescripts.com/forum/thread64728.html

[jek]

the axis is different for a * in COUNT(). it doesn't mean "expand to all columns" like it would in SELECT , it's more "expand to all rows" with no further evaluation whatsoever. it shouldn't ever be slower than COUNT(1). i'd bet lots of dbs rewrite a contant, true argument for COUNT() as COUNT() interally...

[Mike Bayer]

the original thread is at: http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:1156159920245

Its pretty clear that very old versions of oracle (like, version 7 or earlier) had some internal expansion happening with count(*)...that idea is very widespread and a couple of articles mention "these days, count(1) and count(*) are the same". but pointless now, should probably take it out of the docs (i think a lot of examples reference it).

[Michael Trier]

added in b403f156fedf7f1c3da34818f5ebe9debab481d7

[Mike Bayer]

Removing milestone: 0.5.0 (automated comment)