Ability to use custom partition key and primary keys differs from sorting keys for *MergeTree

[antonio-antuan]

• _get_server_info implemented
• declarative engines parameters

[coveralls]

Coverage increased (+1.0%) to 90.577% when pulling f18ad1e on aCLr:feature/declarative_engines into d753192 on xzkostyan:master.

[xzkostyan]

Я обязательно посмотрю это через недельку, как доберусь до компьютера.

[xzkostyan]

Great job.

One question: why we need _get_server_version_info and why it is muted in test_parse_date_types?

[antonio-antuan]

_get_server_version_info appears in my code before that commit. The problem was that DESCRIBE TABLE returns different number of columns in different versions. Don't remember, what was the solution, but _get_server_version_info was needed. Anyway, there is no reason not to retrieve that info.

That method muted for test_parse_date_types because ALL requests to localhost:8123 will return specified response. As you can see, the response a\nDate\n2012-10-25\n is incorrect server version :)

P.S. If you going to make a new release... I'm going to make several new pull requests (joins, engines reflection and maybe something else) in a short time. Maybe you want to wait for it.

[xzkostyan]

It seems that the problem with DESCRIBE TABLE was fixed in this commit: f3e30ea

Take your time with making PRs. There is no need to hurry. I'm planning to make new release by the end of March.

[antonio-antuan]

Yep, I've seen it.
If there is no reason to have that method implemented it can be removed, of sure. But it is the part of SQLA-engines API, so I think that it can exist in that dialect.