Manyy-to-many and an INSERT-query, do i need a temp?

[D-flyer]

Hi there,

Gonna try to explain my problem. In my db model i got a many-to-many relation. So i have 3 tbl's where one joins the other 2.

Now the flow of the front-end:
page1 a <form> wich collect the data for tbl1 --->>> and with hidden fields the data goes to Page2
page2 a <form> wich collect the data for tbl2 --->>> and with hidden fields the data(from this page and page1) goes to a preview page
preview page --->shows all the data and the user can go back or submit the data.

Now what is was wondering:
To join the data in tbl3 i need the tbl1.id before i submit all the data, right???How do i accomplish this??

Tips, suggestions

[KillAllDash9]

Use several queries.

1. Insert data into table 1.
2. Get last ID and keep in a variable..
3. Insert data into table 2.
4. Get last ID and keep in a differetn variable.
5. Insert IDs from the variables into table 3.

[D-flyer]

thnks,

So when a user decide to NOT post his data, say after page1, i need to somehow flush the data out of the db?? I don't like the idea of stuffing my db with incomplete

I was thinking "temp". Get lastest ID and keep it in a temp variable.. or maybe the use of a sessions???

[Karl]

If you've got a many-to-many relationship then you need to re-think your database design so that you eliminate many-to-many relationships between tables. The generally accepted way it to create a table that ties the other two together e.g.

Table1 has many to many with Table2

So you create Table3, which is just composed of the primary key of Table1 and the primary key of Table2, so you create a bridge between the tables storing the relationship and thus avoid a direct many-to-many relationship.

[D-flyer]

Yep,
I understand the principle how to convert many-to-many relations into a db-struture

But the problem i described is one i can't get done. For example, in tble3 i want to correspond tbl1.id with tbl2.id[2] and tbl2.id[3]. And then with the flow i described earlier.

Any more suggestions???

[Karl]

Sorry, my apologies, only skim read your post, looked more at the tables and flow, oops.

One option is to run a cron job every hour that removes 1st stage data that is older than 1 hour and hasn't had 2nd stage data added.

[wwb_99]

What you really need here is a transaction. Just have it rollback on cancel.

WWB