@mfikes there is a bug in the type inference you added for if when using (if (. response -ok) ...). the test is a list with 3 elements and all syms. so it tries to analyze the . as a function. https://github.com/clojure/clojurescript/blob/b38ded99dc0967a48824d55ea644bee86b4eae5b/src/main/clojure/cljs/analyzer.cljc#L1602

since resolve-var pretty much resolves any symbol it resolves and further checks fail. still seems weird that it is trying to resolve .

doesn't exhibit as a bug in CLJS since it blindly accepts all symbols with a . in them and never warns for them

https://clojure.atlassian.net/browse/CLJS-712

Does the above ticket cover it, or do we need a separate ticket?

well the CLJS-712 kinda hides the other issue. so fixing that would cause warnings. might make sense to fix the other issue first

I don’t mind adding a ticket. I can take care of that

yeah just needs a (not= '. (first test)) check added to the other checks I guess

or do other logic in that case

dunno much about the type inference setup

Maybe even ignore anything in https://github.com/clojure/clojurescript/blob/b38ded99dc0967a48824d55ea644bee86b4eae5b/src/main/clojure/cljs/analyzer.cljc#L1327

(specials)

I’ll put together a simple ticket and patch that does that.

yeah that might make sense

^ https://clojure.atlassian.net/browse/CLJS-3163

no-warn is used when analyzing . in that code, so no warnings should be emitted but, it arguably can’t hurt to bail if the first symbol is a special and avoid any add-on issues