When both "class" and "type-func" attributes are present, the parser currently ignores the latter in favour of the former. That's odd because "class" is specified as mandatory in the rnc/rng files [1]. So, by its very nature, the "type-func" attribute should be the preferred way to get the GType when both are present.

I think that, depending on the order of the attributes, this was possible before commit 287ba6b94b290.

This has consequences for Vala [2]. For GTK+ template classes, the Vala compiler ensures that all [GtkChild] class members in Vala have a matching XML object with the right type. However, it doesn't know about "type-func". So, if only the "type-func" attribute is used, then it will fail to compile because it won't know about the XML object. See vala/codegen/valagtkmodule.vala

It seems easier to fix this in GtkBuilder than to teach valac to get the type from the "type-func". But I know little about Vala, so I might be wrong.

Chances are pretty high that nobody uses "class" and "type-func" together, because if they did, their "type-func" was being happily ignored so far. So, it will be nice to have this in gtk3 too.

[1] The parser doesn't enforce that (bug 786931).
[2] https://bugs.webkit.org/show_bug.cgi?id=175937#c16