stdilb: Check for bad type constraints in function types

The parser recognizes the 'is_subtype(V, T)' syntax for constraints,
and of course the new 'V :: T' syntax, but other variants result in an
error message. Up to now, the parser and linter have let badly formed
constraints through, and relied upon Dialyzer to emit warnings.

is_subtype/2 cannot easily be taken out from the parser. Not only
would we need find a way to emit a (linter) warning, but there also
needs to be an option for suppressing the linter warning as
compilation with +warnings_as_errors has to work. (Notice that the
abstract format representation for 'V :: T' is the same as for
'is_subtype(V, T)'.)

This correction was triggered by an email from Robert, and Kostis
created pull request 1214 to provide a fix. However, Kostis' fix
disallowed is_subtype() altogether, which breaks backward
compatibility.

As of Erlang/OTP 19.0 (ticket OTP-11879), the 'is_subtype(V, T)' is no
longer documented.

1 files changed, 1 insertions, 1 deletions

diff --git a/erts/doc/src/absform.xml b/erts/doc/src/absform.xml
index ab00d47425..fe8e3b30e7 100644
--- a/erts/doc/src/absform.xml
+++ b/erts/doc/src/absform.xml
@@ -886,7 +886,7 @@
         Rep(Fc) = <c>[Rep(C_1), ..., Rep(C_k)]</c>.</p>
 
       <list type="bulleted">
-        <item>If C is a constraint <c>is_subtype(V, T)</c> or <c>V :: T</c>,
+        <item>If C is a constraint <c>V :: T</c>,
           where <c>V</c> is a type variable
           and <c>T</c> is a type, then Rep(C) =
           <c>{type,LINE,constraint,[{atom,LINE,is_subtype},[Rep(V),Rep(T)]]}</c>.