Implement astuple method

[Insoleet]

Please checkout my proposal #77. Do not hesitate to comment this code.
The parameter tuple_factory is only tested with the tuple type. If you think this should be tested with other factories, do not hesitate to tell me.

[codecov-io]

Current coverage is 100% (diff: 100%)

Merging #78 into master will not change coverage

@@           master   #78   diff @@
===================================
  Files           8     8          
  Lines         446   466    +20   
  Methods         0     0          
  Messages        0     0          
  Branches       97   105     +8   
===================================
+ Hits          446   466    +20   
  Misses          0     0          
  Partials        0     0          

Powered by Codecov. Last update ca2a1e1...eebe654

[Tinche]

@hynek How do we feel about filter in astuple? My gut feeling is a little iffy since if you skip an element in the tuple, it fundamentally changes the meaning of the tuple. But maybe that's what you want sometimes.

[Tinche]

Also I'm thinking we need either a dict_factory parameter or to have retain_collection_types to apply to dicts too, since right now astuple will map all OrderedDicts to dicts and there should be a way to avoid that.

I think just having retain_collection_types apply in the elif isinstance(v, dict): branch would be enough. (Use v.class as the dict factory.)

[hynek]

I very much think that filter makes sense. Especially in the use case with DB you may want to keep stuff out. Or drop a password field or something.

[Tinche]

Yeah, let's leave it in. I was actually thinking about passwords in the first place. But the exact scenario I was worried about is impossible since no mandatory attributes are allowed after an attribute with a default value, so it's moot anyway.

[Insoleet]

I took all your comments into account.

[Tinche]

Alright, we're getting close for Hynek to take a look. :)
Please add a changelog entry too, you can find examples in CHANGELOG.rst.

[Tinche]

I've noticed retain_collection_type isn't propagated always in astuple. I've expanded our Hypothesis nested class strategy so tests can catch this, and written a test. I don't really feel like pasting it all in here, so let me issue a pull request onto your repository.

[Tinche]

@hynek I think this is in good shape, please take a look when you have a chance.

[hynek]

could you please activate the shiny new maintainer collaboration feature? I don’t want to pester you with minutiae.

[Insoleet]

Done !

[hynek]

I guess this will need some work due to @Tinche’s work on simple_class?

[Tinche]

Yeah, this needs a slight rebase. simple_classes should now be simple_classes(); i.e. the strategy is now parameterized.

[Insoleet]

Ok, I fixed it.

[hynek]

Thank you!