Reputation: 140753
What is the trick in MySql (Version 4) to write from PHP thtree Inserts and to be sure that we can link tables properly.
[Table1]-[LinkTable]-[Table2]
In the PHP code I create an Insert that add 1 row in table 1 and an other Insert that add an other row in table 2. I get both rows PK (primary key) with two SELECTs that return me the last row of each tables. This way I get the PK (auto-increment) from Table 1 and Table 2 and I insert it on the link table.
The problem is that is not atomic and now that I get a lot of transaction something it doesn't link properly.
How can I make a link between 2 tables in MySQL 4 that preserve data? I cannot use any stored procedure.
Upvotes: 1
Views: 2326
Reputation: 105878
Well, if you have the ability to use InnoDB tables (instead of MyISAM) then you can use transactions. Here's some rough code
<?php
// Start a transaction
mysql_query( 'begin' );
// Execute the queries
if ( mysql_query( "insert into table_one (col1, col2) values('hello','world')" ) )
{
$pk1 = mysql_insert_id();
if ( mysql_query( "insert into table_two (col1, col2) values('foo', 'bar')" ) )
{
$pk2 = mysql_insert_id();
$success = mysql_query( "insert into link_table (fk1, fk2) values($pk1, $pk2)" );
}
}
// Complete the transaction
if ( $success )
{
mysql_query( 'commit' );
} else {
mysql_query( 'rollback' );
}
If you can't use InnoDB tables then I suggest looking into the Unit Of Work pattern.
Upvotes: 2
Reputation: 7956
your problem is here:
SELECTs that return me the last row of each tables.
if you did something like this: SELECT MAX(id) FROM table
you can get a wrong id for your transaction.
This is a common many-to-many relationship
as BaileyP say you should use the function mysql_insert_id().
at https://www.php.net/mysql_insert_id
you can read
Retrieves the ID generated for an AUTO_INCREMENT column by the previous INSERT query.
so dont care if other process insert a row in the same table. you always will get the last id for your current connection.
Upvotes: 1