Clojurians Log v2

Continuing my sequence of annoying compiler questions: 1. In TryExpr, why are hasJavaClass and getJavaClass computed from tryExpr only? Catch clauses also can return values; those values do not have to agree with tryExpr. I realize Clojure is not very strict about these things, but for example one might be misled into doing an explicit method call rather than reflection. (`TryExpr` is not a MaybePrimitiveExpr, so at least we don't have to worry about primitives here.) 2. Why is there a tag field in ObjExpr? (And by inheritance, FnExpr and NewInstanceExpr .) The only places it is referenced are ObjExpr.getJavaClass and its override in FnExpr.getJavaClass. During the loading of the entire corpus of Clojure source code, the tag is always null. During the load, the FnExpr.getJavaClass method is called, but in only three contexts: (a) we are checking if it is a primitive type (not) -- this most often occurs when an fn is the binding of a local; (b) we are trying to access the __methodImplCache field (in which case we need to be an AFunction which is the default for FnExpr.getJavaClass when no tag is present), and (c) some case statements in the spec code where the return values in the cases are all fns. Is there some circumstance where the tag would be used? 3. Let's dig in a little deeper. The code for ObjExpr.hasJavaClass is

Class jc;
public Class getJavaClass() {
	if (jc == null)
       jc = (compiledClass != null) ? compiledClass
              : (tag != null) ? HostExpr.tagToClass(tag)
              : IFn.class;
     return jc;
}

public Class getJavaClass() {
        if (jc == null)
            jc = tag != null ? HostExpr.tagToClass(tag) : AFunction.class;
        return jc;
	}

sorry for not replying to the earlier ones, tend to miss stuff on the weekend. I haven't looked into them, not sure when I will have a chance, these kinds of questions usually take some archaeology and often the answer is more a side effect of a series of changes over time rather than intention.