[RFC] remove static initialization of TableMap

[jaugustin]

We need to find a good way to initialize the TableMapclasses and remove the static call
TableMap::buildTableMap() at the bottom of each TableMapclasses

Maybe we could use :

• composer class mapping (to auto discover classes)
• use the propel container
• ...

[willdurand]

Can we live with these static calls by now?

[deadbeef84]

What about using some sort of singleton-pattern for the TableMap-classes? Are there any reasons to create multiple instances?

Another problem with the static calls is that you get different results in the DatabaseMap depending on what classes have been loaded. Would it be possible to let the DatabaseMap know about the tables when loading propel configuration?

[hhamon]

A factory object to create and initialize the corresponding Table map?

[fzaninotto]

Bike shedding imho.

Le 11 mai 2013 à 11:55, William Durand notifications@github.com a écrit :

Can we live with these static calls by now?

—
Reply to this email directly or view it on GitHub.

[jaugustin]

@willdurand yes we can but it would be better to remove it before beta otherwise we are going to keep those lines forever ;)

@deadbeef84 I think we can generate a classmap with the builder and give it to the tablemap with propel configuration

[marcj]

Generating a DatabaseMap for each database used in the schemas could solve that. E.g. \DatabaseMap\BookstoreMap extends \DatabaseMap\Base\BookstoreMap that contains then relations to all containing tables and can act as a factory.

[NobleUplift]

@fzaninotto This is definitely not bikeshedding. When used incorrectly TableMaps can blow out memory.

[dereuromark]

@mringler Can we close this with latest changes done?

[mringler]

Yes, we handled this. Let's close the issue.