Script to test ORM vs core + fixes

[MatBarba]

• Create a simple test script (to run manually) to test load all the tables defined in the ORM from a real core database, to flag any that fail to load
• Fix a few table definitions by removing an extra primary key in the *_attrib tables, and replacing them with a primary_key parameter on the columns the core schema use as unique
• NB: I also had to fix the string length to a definite value String(500), because SQLAlchemy needs that to create a primary key constraint in MySQL. This works now, but it might be problematic if we want to use the ORM to load very long attribs into a database. The best way to handle this really would be to alter the core SQL schema to have an explicit primary key as a simple autoincremented row ID.

[JAlvarezJarreta]

Looks good, and thank you for the new test, quite useful.

[JAlvarezJarreta]

Re. the primary key and Text, that restriction actually makes sense, and the core schema "is wrong": it does not make sense to have a primary key of 10000 characters. Something to fix in the new world IMO.

[ens-LCampbell]

Approved, just wanted to ask if the changes to models will now be in place but still requires changes to the scehma itself to conform ? How does that affect models utility in the interim ?

[JAlvarezJarreta]

"URL vs server details" discussion left, otherwise looks good 👍

[MatBarba]

Approved, just wanted to ask if the changes to models will now be in place but still requires changes to the scehma itself to conform ? How does that affect models utility in the interim ?

Right now the SQL schema from ensembl should be compatible with the ORM from ensembl-py, but with the possibility that we would not be able to handle long value for the *_attrib tables.

If we apply my recommendation to add an explicit key to all *_attrib tables, then we would just need to change the value fields back to a Text type, and define the primary_key=True only for the new row.