Hi,
the difference is how the value of the variable is used. A type
expression does not have a covered text, so the type is just some
something that point to some annotation(s) in a local context. In the
first example, the "rootGovernor" type variables is resolved against one
annotation using different "lookup" strategies, e.g., first it looks for
annotations of the given type with the same offsets as the matches rule
element. It tries to guess which annotation you meant. So it all depends
on the value of the variable. Without testing and looking at the code,
I'd assume that the assignment is on the type of the feature value of
the last DEP_ROOT annotation in the document. And only if the expression
is correctly resolved as an AnnotationTypeExpression. It's hard to test
it without an specific input example.
In the second example, nothing is resolved against the matched
annotation, but only the value of the variable is logged. The value is
determined by the last match of the previous rule.
Try to declare "rootGovernor" as ANNOTATION instead of TYPE to see the
difference.
Best,
Peter
Am 14.06.2017 um 14:34 schrieb Reza Hay:
Hi All,
It seems if a global variable does contain a type instead of a primitive value,
its scope is limited to the current statement. Example:
STRING s;
TYPE rootGovernor;
(DEP_ROOT & VERB) {-> rootGovernor = DEP_ROOT.Governor, LOG("root: " +
rootGovernor.ct)};
ANY{->LOG("root: " + rootGovernor.ct)};
The first statement prints the correct answer but the second statement prints
null. However, in
(DEP_ROOT & VERB) {-> s = DEP_ROOT.Governor.ct};
ANY{->LOG("root: " + s)};
the variable s has still its value and the second statement prints the expected
value. Any idea?
Regards,
Reza
