Support to dump and load of postgresql schemas

[lucasts]

cc: @jonleighton, @tenderlove

As of PR #1410, this change try to respect table with qualified schemas in postgres

now you can do something like this on a migration file:

    create_table 'myschema.mytable' do |t|  
      t.string :column          , :null => false, :limit => 20
      t.timestamps  
    end

something I want to work soon is to create schemas in migrations without breaking dump and create tasks(that will also break test tasks)

[jonleighton]

Hiya,

This looks identical to what I reverted, which broke the tests.

The reason it broke is because this line is checking for 'schema_things', but after your change the entry is 'rails_pg_schema_user1.schema_things', hence the assertion fails.

I'm not sure what the 'right' behaviour is here so I will let @tenderlove comment as he is more familiar with the adapters code.

Jon

[lucasts]

Hi Jon,

Yes, almost the same.
The fix is at line 663 https://github.com/rails/rails/pull/1450/files#L0R663
Now I'm checking if the schemaname is a authorization schema for the current user.

I ran the test that you're pointing and it pass now.

Did you run it with this new patch?

[jonleighton]

Ok, I didn't notice the difference. I will let @tenderlove check it over.

Just on a code quality point, I would prefer if you hadn't changed the if/else statement into a ternary.

[lucasts]

Ok, lets wait for him.
About your point, I stopped for 5min to think about it before rewrote as ternary.
Is that preference (if/else over ternary) a ruby pattern?

thanks for the tip.

[jonleighton]

There's no hard and fast rule, but if an if/else is at all complicated, then I favour a full if/else rather than ternary. I find it easier to read. Especially in this case where your condition includes an == in addition to the = in the assignment.

[lucasts]

Agreed @jonleighton
The multiple '=' chars in the line may lead to misunderstandings in the future.

[lucasts]

@tenderlove,
As the "same" PR was merged before, but reverted due to a little break, that I fixed now

Is anything else left to merge this?

Cheers
-lucas

[isaacsanders]

Is this still an issue @lucasts?

[rafaelfranca]

This pull request cannot be automatically merged. If you still want to merge it, please rebase against the master. If no, please close it.

Thanks.

[lucasts]

I'll close this PR and investigate the current status of the root problem.

If I found something that need fix I will create a new PR with uptated code. Since last interaction I found other issues with postgresql schema support that need to be fixed too.