Re: percentile value check can be slow

2017-11-19 Thread jotpe
On 19.11.2017 18:49, David Fetter wrote: On Sun, Nov 19, 2017 at 01:23:42PM +0100, Tomas Vondra wrote: Hi, On 11/19/2017 03:10 AM, David Fetter wrote: On Sat, Nov 18, 2017 at 11:05:47PM +0100, Tomas Vondra wrote: Hi, ... Is 'recognizer' an established definition I should know? Is it the s

Re: percentile value check can be slow

2017-11-19 Thread David Fetter
On Sun, Nov 19, 2017 at 01:23:42PM +0100, Tomas Vondra wrote: > Hi, > > On 11/19/2017 03:10 AM, David Fetter wrote: > > On Sat, Nov 18, 2017 at 11:05:47PM +0100, Tomas Vondra wrote: > >> Hi, > >> > >> ... > >> > >> Is 'recognizer' an established definition I should know? Is it the same > >> as 'va

Re: percentile value check can be slow

2017-11-19 Thread jotpe
On 19.11.2017 13:23, Tomas Vondra wrote: Hi, On 11/19/2017 03:10 AM, David Fetter wrote: On Sat, Nov 18, 2017 at 11:05:47PM +0100, Tomas Vondra wrote: Hi, ... Is 'recognizer' an established definition I should know? Is it the same as 'validator' or is it something new/different? I borrow

Re: percentile value check can be slow

2017-11-19 Thread Tomas Vondra
Hi, On 11/19/2017 03:10 AM, David Fetter wrote: > On Sat, Nov 18, 2017 at 11:05:47PM +0100, Tomas Vondra wrote: >> Hi, >> >> ... >> >> Is 'recognizer' an established definition I should know? Is it the same >> as 'validator' or is it something new/different? > > I borrowed it from http://langsec.

Re: percentile value check can be slow

2017-11-18 Thread David Fetter
On Sat, Nov 18, 2017 at 11:05:47PM +0100, Tomas Vondra wrote: > Hi, > > On 11/18/2017 10:30 PM, David Fetter wrote: > > On Sat, Nov 18, 2017 at 10:44:36AM -0500, Tom Lane wrote: > >> jotpe writes: > >>> I tried to enter invalid percentile fractions, and was astonished > >>> that it seems to be ch

Re: percentile value check can be slow

2017-11-18 Thread Tomas Vondra
Hi, On 11/18/2017 10:30 PM, David Fetter wrote: > On Sat, Nov 18, 2017 at 10:44:36AM -0500, Tom Lane wrote: >> jotpe writes: >>> I tried to enter invalid percentile fractions, and was astonished >>> that it seems to be checked after many work is done? >> >> IIRC, only the aggregate's final-functi

Re: percentile value check can be slow

2017-11-18 Thread David Fetter
On Sat, Nov 18, 2017 at 10:44:36AM -0500, Tom Lane wrote: > jotpe writes: > > I tried to enter invalid percentile fractions, and was astonished > > that it seems to be checked after many work is done? > > IIRC, only the aggregate's final-function is concerned with direct > arguments, so it's not

Re: percentile value check can be slow

2017-11-18 Thread Tom Lane
jotpe writes: > I tried to enter invalid percentile fractions, and was astonished that > it seems to be checked after many work is done? IIRC, only the aggregate's final-function is concerned with direct arguments, so it's not all that astonishing. regards, tom lane