The issue with the static access operator (the dot)

[OvermindDL1]

@Qqwy

Anyhow, I like Foo.bar(my_foo) a lot better than my_foo.bar, because it is more explicit and pipeable.

Would be in mine as well, instead of thing.vwoop(2) you can do MyTesting.vwoop(2, thing) (I’m using Erlang functionality, which puts the thing into the last slot as it should be, unlike Elixir’s weird way, and I cannot fix that, however I could swap it if it is detected in the last slot with ease though in the macro case generator).

[ericmj]

@Qqwy

One thing that static typing nor Dialyzer would fix, is what happens when you write some code, publish it as a library, and then people start matching on your struct’s structure.

Because of backwards-compatibility, the structure of your struct is now ‘frozen’. The old fields will need to be kept in there because some other code might depend on it.

Typespecs to the rescue. Declare your types as opaque and users should know they shouldn’t depend on the structure. If they use dialyzer they will even get errors when matching on the structure or using the dot field access.

[DianaOlympos]

BTW this is exactly how Liskov and Zilles handle types in their “Programming with Abstract Data Types” paper. So seems it has some reasons

[Qqwy]

@ericmj

@Qqwy

One thing that static typing nor Dialyzer would fix, is what happens when you write some code, publish it as a library, and then people start matching on your struct’s structure.

Because of backwards-compatibility, the structure of your struct is now ‘frozen’. The old fields will need to be kept in there because some other code might depend on it.

Typespecs to the rescue. Declare your types as opaque and users should know they shouldn’t depend on the structure. If they use dialyzer they will even get errors when matching on the structure or using the dot field access.

Yes, exactly! I feel like more library designers should be aware of this option, and should use this, in my opinion.

[OvermindDL1]

@ericmj

Typespecs to the rescue. Declare your types as opaque and users should know they shouldn’t depend on the structure. If they use dialyzer they will even get errors when matching on the structure or using the dot field access.

Exactly what I said, this is why -opaque exists (er, @opaque in Elixir). ^.^