CREATE TABLE `test` (
`id` int(11) NOT NULL,
`md5` char(32) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `md5` (`md5`)
) ENGINE=InnoDB

INSERT IGNORE INTO test (md5) values (‘A’)

1 row affected

INSERT IGNORE INTO test (md5) values (‘A’)

Duplicate entry ‘A’ for key ‘md5′

I think an incredibly useful feature here would be to return the innodb row id of the row which generated the duplicate key error. This would save us a select statement (either before or after the insert) to determine which row caused the dupe error. (I guess the unique row_id would not be of much help, so only return the value of PRIMARY if it is defined)

Is there something existing like this already? I dumped the mysqli connection object in php, and I see nothing like that. Insert_id of course, but that can only be used when there is an auto_increment column.

Someone will answer (why not just use on duplicate key update?) Well this is because i want “on duplicate key return unique_row_id”

Cheers

//Steve

field2= CASE WHEN field2’$field1′ THEN ‘$field3′ ELSE field3 // this does not work off the existing (old) field1 value!

field1= CASE WHEN field1>’$field1′ THEN ‘$field1′ ELSE field1 // this gets a low value
field2= CASE WHEN field2′$field1′ THEN ‘$field3′ ELSE field3 // this does not work off the existing (old) field1 value!

This should update field3 when field1 changes. But it does not.

field3= CASE WHEN field1=’$field1′ THEN ‘$field3′ ELSE field3 // if field3 is time you will only see the most recent value match of field1

This does work but it looks to see if field1 is equal to the new value.
The problem here is that I can only have the last match, I would like to have the first match in some cases.

It seems that old and new values are only available within a given field. Is that so?

Is there another way to have old/existing values available from one field in the case switch or value of another field.

Thanks.

how to do like this

insert into (field1, field2, field3) values (1,2,3),(1,2,3) and so including on duplicate update. because i want insert/update many records from am json source.
This should have been posted in a forum but since it deals my problem exactly i thought of posting it here.

In order for this to work, the PRIMARY KEY defined for the table should only contain one column–or if it has multiple columns–the data you’re going to insert/update for the given columns should be identical to the columns defined in the PRIMARY KEY.

To illustrate:

PRIMARY_KEY(`id`, `timestamp`) == a single key (but with two values, thus more unique)

The `id` is very likely always going to be unique. However, adding the `timestamp` to the primary key makes it even more unique; so, if the `id` exists in the table and you simply want to update the `timestamp`, the new value for the timestamp you’re providing will mean that `id` matches but `timestamp` does not, because you’ve defined the primary key to be `id` and `timestamp`, which means BOTH values you’re inserting/updating must equal the values currently in the database to make the primary key match and therefore update the row instead of insert a new row with the same `id`, but different `timestamp`.

I use insert into ad_stat (banner_id, date, views) values ($banner_id, ‘$date’, 1) ON DUPLICATE KEY UPDATE views=views+1

So this query firstly tries to insert and only then to Update – is there opposite way like:
UPDATE … ON NO ROWS INSERT ???
Because there are 5 inserts daily and 50 000 updates – wouldn’t it somehow increase performance?

I have a dilema with INSERT (fld1, fld2, fld3) VALUES(’1′,’1a’,’1b’),(’2′,’2a’,’2b’),(’3′,’3a’,’3b’) ON DUPLICATE KEY UPDATE fld2 = ??? ( I want this to be the value that would be inserted ie ’2a’) (fld1 is a unique index)

What should go in the ??? spot?

Thanks.