the value grows exponentially with LOC (lines of code) count. :)
Exponentially? Now I'm missing something... Your way has 156 chars: data SomeDataType = SomeDataType { sdtField1 :: SDT_Field1Type, sdtField2 ::
SDT_Field2Type, sdtField3 :: SDT_Field2Type, sdtField3 = f sdtField1 sdtField2}
This way has 162 chars: data SomeDataType = SomeDataType { sdtField1 :: SDT_Field1Type, sdtField2 :: SDT_Field2Type} sdtField3 :: SDT_Field2Type sdtField3 = f <$> sdtField1 <*> sdtField2 This adds 6 characters per dependent deconstructor function. As a fraction of total LOC, this is insignificant. Belka wrote:
Thank you, for your reply, Dan! :)
You don't really need this inline in the record syntax, do you? In fact, that was the point. To enclose direct functional dependants into the record declaration. To achieve better pithiness - it's valuable, and the value grows exponentially with LOC (lines of code) count. :)
sdtField3 sdt = f <$> sdtField1 <*> sdtField2 Doesn't look much better than my "under" function (t `under` f = \x y -> (x f) `t` (y f)). What did I miss? I believe, there are good reasons to use Control.Applicative for lots purposes, but unfortunately, yet haven't had time to try it in my practice.
Belka