Support PostgreSQL materialized views

[kastiglione]

The change is pretty minimal, the query used in #table_exists? has been expanded to include materialized views in the kinds of relations it includes in its search.

In order to share the existing view tests, I made the previous test class into a concern that is then included into separate ViewTest and MaterializedViewTest classes, the latter only if PostgreSQL is at least version 9.3. The test class then uses the name of the test to determine the type of view to create and test against.

[kastiglione]

I've had a look at the Travis logs, but there isn't much to go off. Is there a way to find out info about whichever error was raised?

[rafaelfranca]

Why do we need to reopen the same module in the same file?

[kastiglione]

I did that purely for top-down readability. If I define ViewTestConcern entirely up front, that pushes the two test class definitions to the bottom of the file. I thought it would be preferable to put them at the top. So no technical reason, it can be changed.

[kastiglione]

Note that in dd3aba0 I did a bit of clean-up by removing #extract_pg_identifier_from_name which duplicates the behavior of Utils.extract_schema_and_table. The former was used in just one method, and has no tests while the latter does have tests.

[kastiglione]

😎

Apologies if it's bad form to bump a PR, I'm just unsure that this won't get lost in the crowd.

[senny]

@kastiglione Will work on this once 4.1 is out. I assigned the 4.2 milestone.

[kastiglione]

✨ thanks @senny

[senny]

@kastiglione 4-1-stable is forked off and we can move this forward. After skimming through the changes I noticed there are quite a bit of unrelated changes. Can you cherry-pick the refactoring changes onto another branch and submit a PR from there? I'd like to keep this one focused on adding support for materialized views only.

The refactorings might have a different impact. For example it favors the slower extract_schema_and_table method. See this gist. We should evaluate these in a separate PR.

[kastiglione]

Sure, sounds good. Was trying to clean up what I saw as obvious duplication, but indeed they are separate changes.

[senny]

@kastiglione ping, wanted to check back if you have time to work on this? Please let me know if you don't so I can move this forward. Thank you for your work 💛

[kastiglione]

I don't have much time, but also work impediments besides time. Can I simply revert the last two commits, the ones that mucked with schema extraction? The first three commits were the ones core to supporting materialized views.

Thanks to you as well.

[senny]

yes, you can. Make sure to squash the other commits into a single one and rebase your branch. It does no longer apply.

[senny]

I squashed the commits and applied the patch via cherry-pick (def6071)

@kastiglione thank you for your work!

[kastiglione]

Thanks @senny. My apologies for any delay I might have caused.

[senny]

@kastiglione no worries, we have plenty of time until 4.2 hits 😄