"unification surprise" with morphology

Related to this but may be separate.

I think I am maybe seeing one of those “unification surprises” where you don’t see an edge in the chart but aren’t getting any unification failures interactively.

Here’s what I expect (the verb):

Screen Shot 2020-09-18 at 1.32.06 PM

The order of lexical rules is: no-aspect, non-3rd-pl, non-future, and finally interrogative-3sg-transitive.

I am no longer getting this with the updated customization system/grammar, and specifically what I am not getting is the non-future—interrogative-transitive part.

I am seeing the interrogative-IN-transitive rule in the chart but not the transitive rule. Yet when I am interactively unifying, everything seems fine.

Which item on the “surprise list” does this most sound like?

It is a const-lex-rule, and the difference between the working version and the broken version is that the broken version inherits additionally from something called add-icons-subj-foc-lex-rule, which is something new, being added by the information structure library. The intransitive version is not inheriting from it, and if I remove it and replace it with add-only-no-rels-hcons-rule, then I do get the parse.

; I get this in the chart:

ITRG-3SG-INTRAN-lex-rule := ITRG3PER-lex-rule-super & add-only-no-rels-hcons-rule & itrg-lex-rule &
  [ DTR.ARG-ST #arg-st,
    ARG-ST #arg-st &
           < [ ] >,
    SYNSEM.LOCAL.CAT.VAL.SUBJ.FIRST.LOCAL [ CONT.HOOK [ INDEX.PNG [ PER 3rd,
                                                                    NUM SG ],
                                                        ICONS-KEY non-focus ],
                                            CAT.HEAD.CASE Nom ] ].

; I do not get this in the chart:

ITRG-3SG-TRAN-lex-rule := ITRG3PER-lex-rule-super & add-icons-subj-foc-lex-rule & itrg-lex-rule &
  [ DTR.ARG-ST #arg-st,
    ARG-ST #arg-st &
           < [ ],
             [ ] >,
    SYNSEM.LOCAL.CAT.VAL [ SUBJ.FIRST.LOCAL [ CAT.HEAD.CASE Nom,
                                              CONT.HOOK [ INDEX.PNG [ NUM SG,
                                                                      PER 3rd ],
                                                          ICONS-KEY focus ] ],
                           COMPS.FIRST.LOCAL [ CONT.HOOK.ICONS-KEY non-focus,
                                               CAT.HEAD.CASE NFO+Acc ] ] ].

The new information structure rule which seems to be breaking things looks like this:

add-icons-subj-foc-lex-rule := add-only-no-rels-hcons-rule &
  [ SYNSEM.LOCAL [ CAT.VAL.SUBJ < [ LOCAL.CONT.HOOK.INDEX #target ] >,
                   CONT [ HOOK.INDEX #clause,
                          ICONS.LIST < focus &
                                       [ IARG1 #clause,
                                         IARG2 #target ] > ] ] ].

The interrogative supertype looks like this:

itrg-lex-rule := add-only-no-ccont-rule &
  [ SYNSEM.LOCAL.CONT.HOOK.INDEX.SF ques ].

Why would the combination of all that result in the transitive rule not showing up in the chart?

(By the way, if I get rid of those new information structure rules, the test just passes… Given that those new rules were just experimental and only covered two of the corresponding information structure tests, I am half-mind just skipping those two tests and getting rid of them. But, would still be good to understand what is going on, and maybe it is an easy fix.)