Re: Serious LEFT JOIN bug in 3.23

2001-03-21 Thread Benjamin Pflugmann
On Wed, Mar 21, 2001 at 01:32:11AM -0700, [EMAIL PROTECTED] wrote: > I appreciate your explanation. You may be correct about Standard SQL, > however I do not believe that this can be explained as a simple > cross-product. If a LEFT JOIN were implemented starting with a cross product > then NO LEFT

Re: Serious LEFT JOIN bug in 3.23

2001-03-21 Thread Michael Griffith
I appreciate your explanation. You may be correct about Standard SQL, however I do not believe that this can be explained as a simple cross-product. If a LEFT JOIN were implemented starting with a cross product then NO LEFT JOIN would EVER return NULL in the right table. A pure cross product would

Re: Serious LEFT JOIN bug in 3.23

2001-03-21 Thread Benjamin Pflugmann
Hi. Ah okay, I as thought, I misunderstood your objection. The behaviour is how Standard SQL defines it and what is to be expected. You may want consider to do some reading on relational database theory (don't take this proposal wrong...). Aside from that, I agree, that sometimes the behaviour

Re: Serious LEFT JOIN bug in 3.23

2001-03-21 Thread Michael Griffith
The query returns 2 different result sets depending on whether the table on the right side of a LEFT JOIN has records or not. I just don't see how this is acceptable behavior. Either it should always return the result set or it should always return the empty set. The size of the right table should

Re: Serious LEFT JOIN bug in 3.23

2001-03-21 Thread Benjamin Pflugmann
Hi. On Mon, Mar 19, 2001 at 09:06:49PM -0700, [EMAIL PROTECTED] wrote: > > hist.id will never be NULL and 5 at the same time. Your clauses conflict > > with each other. > > You need to re-write your query. > > This is ture, except the query suceeds if there is records in 'hist' > Even if it is

Re: Serious LEFT JOIN bug in 3.23

2001-03-19 Thread Michael Griffith
> hist.id will never be NULL and 5 at the same time. Your clauses conflict > with each other. > You need to re-write your query. This is ture, except the query suceeds if there is records in 'hist' Even if it is not a bug it is at least inconsistent behaviour of MySQL > > "Michael Griffith" <

Re: Serious LEFT JOIN bug in 3.23

2001-03-19 Thread btjones
hist.id will never be NULL and 5 at the same time. Your clauses conflict with each other. You need to re-write your query. "Michael Griffith" <[EMAIL PROTECTED]> wrote: EXPLAIN LEFT JOIN fails when joining on an empty table with constant in WHERE clause CREATE TABLE idx ( id mediumint,

Serious LEFT JOIN bug in 3.23

2001-03-19 Thread Michael Griffith
EXPLAIN LEFT JOIN fails when joining on an empty table with constant in WHERE clause CREATE TABLE idx ( id mediumint, KEY id (id), ) TYPE=MyISAM CREATE TABLE hist ( id mediumint, link mediumint, KEY id (id,link) ) TYPE=MyISAM Table idx is populated with thousands or records. Consider