Remove nothrow and data from the 1.4 grammar

[mandolaerik]

No description provided.

[mandolaerik]

urgh, I get a failure if I apply this:

--- a/test/1.2/misc/porting.dml
+++ b/test/1.2/misc/porting.dml
@@ -416,7 +416,7 @@ template evt {
     method get_event_info(void *data) -> (attr_value_t attr) {
         attr = SIM_make_attr_nil();
     }
-    method set_event_info(attr_value_t attr) -> (void *data) {
+    method set_event_info(attr_value_t attr) -> (void *session) {
     }
     method destroy(void *data) {}
     method event(void *data) {}
diff --git a/test/1.4/misc/porting.dml b/test/1.4/misc/porting.dml
index d86cd81b..7f422998 100644
--- a/test/1.4/misc/porting.dml
+++ b/test/1.4/misc/porting.dml
@@ -431,8 +431,8 @@ template evt is event {
         return SIM_make_attr_nil();
     }
     method set_event_info(attr_value_t attr) -> (void *) {
-        local void *data;
-        return data;
+        local void *session;
+        return session;
     }
     method destroy(void *data) {}
     method event(void *data) {}

[mandolaerik]

Never mind, session fails because it is not a valid variable name in 1.4, so that's not a relevant use case (port-dml will create broken DML, but that's ok). Changed outarg name from data to param to cover the same corner of the port_dml script.

[syssimics]

PR Verification: ✅ success

[lwaern-intel]

wait so the old code failed to properly have PARAMETER as part of reserved_idents? That's hilarious. Given how DML 1.2 code has clearly adapted to this, I'd vote for keeping parameter restricted.

[mandolaerik]

I vote against that because it complicates code for no gain. The gain of keeping it reserved is to open for extending syntax in some ways, like allowing parameter declarations inside methods, and if we ever will want to do that kind of change, then it will be 1.4 exclusive.

[syssimics]

PR Verification: ✅ success