Hartmut Holzgraefe wrote:
> Michael Widenius wrote:
>> Hartmut, what information are you missing ?
>>
>> The above is as quite good way to describe it without going to pseudo
>> code.
>>
>> The only things I can see that could be added are:
>>
>> - Indexes are numbered from 1 uppwards, in the same order as given
>> by SHOW CREATE TABLE or SHOW INDEX
>>
>> - The index map is a bitmap that shows which index we will use for
>> checking if we can use an index to retrieve the rows with either the
>> range or index_merge methods. Ie, the value 5 (= 1 + 4 ) would
>> tell you that mysqld will try to use index 1 and 3 for this.
>
> that is exactly what was missing: the explanation of "N"
> in the "(index map: N)" part ... which lead to all the
> guessing (both right and wrong) in this thread ...
>
Now I'm MORE confused. Is this a typo? My interpretation of the value
5 is indexes one and four are used, not one and three. From
http://dev.mysql.com/sources/doxygen/mysql-5.1/sql__select_8cc-source.html:
14452 {
14453 char buf[MAX_KEY/8+1];
14454 extra.append(STRING_WITH_LEN("; Range checked for each "
14455 "record (index map: 0x"));
14456 extra.append(tab->keys.print(buf));
14457 extra.append(')');
14458 }
tab->keys seems to be a Bitmap< 64 >, if the Doxygen-ized source is
linked right.
I also think the index map might be redundant to possible_keys. I
haven't found a case where they differ, though they don't seem to come
from the same code. Can anyone confirm whether this is true? Perhaps
the index map is a subset (but not a proper subset) of possible_keys?
Thanks
Baron
PS I'm not sure this is 'urgent' for Monty -- it's not urgent for me --
but I appreciate everyone's time very much.
