diff options
author | Stephane Glondu <steph@glondu.net> | 2012-01-12 16:02:20 +0100 |
---|---|---|
committer | Stephane Glondu <steph@glondu.net> | 2012-01-12 16:02:20 +0100 |
commit | 97fefe1fcca363a1317e066e7f4b99b9c1e9987b (patch) | |
tree | 97ec6b7d831cc5fb66328b0c63a11db1cbb2f158 /test-suite/success/universes-coercion.v | |
parent | 300293c119981054c95182a90c829058530a6b6f (diff) |
Imported Upstream version 8.4~betaupstream/8.4_beta
Diffstat (limited to 'test-suite/success/universes-coercion.v')
-rw-r--r-- | test-suite/success/universes-coercion.v | 22 |
1 files changed, 22 insertions, 0 deletions
diff --git a/test-suite/success/universes-coercion.v b/test-suite/success/universes-coercion.v new file mode 100644 index 00000000..d7504340 --- /dev/null +++ b/test-suite/success/universes-coercion.v @@ -0,0 +1,22 @@ +(* This example used to emphasize the absence of LEGO-style universe + polymorphism; Matthieu's improvements of typing on 2011/3/11 now + makes (apparently) that Amokrane's automatic eta-expansion in the + coercion mechanism works; this makes its illustration as a "weakness" + of universe polymorphism obsolete (example submitted by Randy Pollack). + + Note that this example is not an evidence that the current + non-kernel eta-expansion behavior is the most expected one. +*) + +Parameter K : forall T : Type, T -> T. +Check (K (forall T : Type, T -> T) K). + +(* + note that the inferred term is + "(K (forall T (* u1 *) : Type, T -> T) (fun T:Type (* u1 *) => K T))" + which is not eta-equivalent to + "(K (forall T : Type, T -> T) K" + because the eta-expansion of the latter + "(K (forall T : Type, T -> T) (fun T:Type (* u2 *) => K T)" + assuming K of type "forall T (* u2 *) : Type, T -> T" +*) |