<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
</head>
<body bgcolor="#FFFFFF" text="#000000">
The definition of "strictly positive" for data types is given <a
href="https://agda.readthedocs.io/en/latest/language/data-types.html#strict-positivity">here</a>,
but apparently it's different for record types, ruling out *any*
occurrence to the left of an arrow (not just those in negative
positions). I doubt that allowing such occurrences could lead to an
inconsistency. What's the rationale for being so strict about
positivity?<br>
<br>
--okay<br>
data D : Set₁ where<br>
d : D → (Set → D) → D<br>
<br>
--fails<br>
record R : Set₁ where<br>
inductive<br>
field<br>
r : (Set → R) → R<br>
<br>
--R is not strictly positive, because it occurs<br>
--to the left of an arrow<br>
--in the definition of R.<br>
<br>
</body>
</html>