On Mar 5, 2014, at 12:40 AM, Luca Bruno <lethalma...@gmail.com> wrote:
> Thanks for sharing. The girparser has code for detecting such cases and
> discard the get_foo for the foo property. It's not something that should go
> in the metadata. Girparser should checked too understand the reason why it
> doesn't discard get_foo in this case.
Thank you for the help. I'm suspicious of the following code snippet in
valagirparser.vala starting at line 920:
prop.set_attribute ("NoAccessorMethod", false);
if (prop.get_accessor != null) {
var m = getter != null ? getter.symbol as Method : null;
// ensure getter vfunc if the property is abstract
if (m != null && (m.is_abstract || m.is_virtual || !prop.is_abstract)) {
getter.process (parser);
if (m.return_type is VoidType || m.get_parameters().size != 0) {
prop.set_attribute ("NoAccessorMethod", true);
} else {
if (getter.name == name) {
foreach (var node in colliding) {
if (node.symbol is Method) {
node.merged = true;
}
}
}
prop.get_accessor.value_type.value_owned =
m.return_type.value_owned;
}
} else {
prop.set_attribute ("NoAccessorMethod", true);
}
}
In the if (getter.name == name), it compares a method name to a property name.
In my test case, I get ("get_enabled" == "enabled"). I added test code and set
getter.merged to true, and the vapi output looks correct - the getter method is
removed. Is this the code that is supposed to remove the getters from the vapi?
If so, the logic does not seem correct to me.
Cheers,
Ed
_______________________________________________
vala-list mailing list
vala-list@gnome.org
https://mail.gnome.org/mailman/listinfo/vala-list
