Infer sort_direction from query

[bernardd]

This change automates detection of the sort_direction field based on the sort ordering in the query. It also enables per-field sort direction, as supported by #34, but does so without the need to explicitly specify them. I've taken the extra tests from #34 and included them in this change (with the addition of explicitly specified amount values on all test data, since the randomly generated values were causing inconsistent test results with those new tests).

[Nickforall]

This is a duplicate of #41, which in its place is also blocked by #34.

I am also sure we do not want to remove sort_direction entirely. I have a couple of production cases where my order-bys are not the same as the sort_direction.

(with the addition of explicitly specified amount values on all test data, since the randomly generated values were causing inconsistent test results with those new tests).

The reason why this was happening in your test results is because I made sure that my random values where always higher than 10 for amount, thus not clashing with the tests asserting ascending sort directions on low values. Though having more reliable amount values may be better here.

paginator/test/support/factory.ex

Line 25 in 31c3ffc

amount: :rand.uniform(100) + 10,

[bernardd]

This is a duplicate of #41, which in its place is also blocked by #34.

Ah dangit. Got confused by the title of #41 and didn't look at it closely. Okay, so if those both work and are ready to go, what do we need to do to unstick them and get them merged? If it's just docs, I'll put together some myself tomorrow (already night here).

The reason why this was happening in your test results is because I made sure that my random values where always higher than 10 for amount, thus not clashing with the tests asserting ascending sort directions on low values. Though having more reliable amount values may be better here.

Yeah, I figured that out. I'm not bothered which solution gets used - random + 10 is just as good to my mind.

[Nickforall]

If it's just docs, I'll put together some myself tomorrow (already night here).

It's just docs afaik, and I haven't heard from @stevedomin since then.
It'd be great to get this going, as I have some other features ready we use in an internal fork, and they are all blocked by #34.

[bernardd]

@Nickforall It might be time to accept that @stevedomin won't be back, at least for a while :) If you have an internal branch all set up and ready to go, how would you feel about publishing it? I'd rather not have multiple of us all working on the same stuff on different repos :)

[bernardd]

Going to close this and focus on trying to get #41 merged