Hah, at Amperity they are remarking at how quickly @dnolen squashed http://dev.clojure.org/jira/browse/CLJS-1739 Rock on!

@dnolen: Wanted to let you know I’m working on a minimal repro without downstream tooling, for a strange regression caused by the 9-element map literal fix. Here is what I know so far, and I’ll of course file a formal JIRA if I sort it https://github.com/mfikes/planck/issues/353

@mfikes hrm seems very strange

I haven’t tracked down the root cause, but I can see that I end up with a KeySeq that has an ArrayNodeSeq inside it, and when -rest is called on the KeySeq, (next mseq) yields () (instead of nil) which leads to KeySeq violating its invariant that it have at least one thing in it, and -first assumes that invariant holds, and it blows up. (I wish I could find a simple form to type in the REPL to repro.)

My (unsubstantiated) hunch is that create-inode-seq and create-array-node-seq are supposed to return either non-empty seqs or nil, and some other code is depending on this.