Robert Lechte

@eradman still haven't implemented an ignore option but have implemented partitioning support so you may not need to filter partitioned tables out manually any longer.

Hey, thanks asking about this. I've been meaning to document solutions in this area in a bit more detail on the docs site - it can work well but there...

> Seems to me like it would be (a lot?) more straight forward if the diff generated by migra would be idempotent itself, not sure if that is possible in...

@harrybiddle if you need data changes you simply add those to your auto-generated script as necessary. If you are managing a single database, then often these scripts are one-offs and...

Yep, I agree that `normalise` is more accurate as a description

Yep, -ize definitely the preferred spelling

Yes migra works with actual connections, not dump files. But you can just create two temporary databases, load the files, and compare them once loaded. [In the docs](http://migra.djrobstep.com/dev_usage/) there's some...

Thanks for filing this issue! In the interests of narrowing down the cause, are you able to share a sanitized snippet of the code you're using to create the connection?

Thank you kindly for this PR! Definitely a much-needed addition to this library. Some things will need changing before we merge: - The current implementation is vulnerable to SQL injection...

Yep I think to work around this, we ought not to have enums from extensions included in the enum list. Perhaps a good solution is a separate dict of `extension_enums`....