01:09 <InPhase> teepee: Do you know off the top of your head if we have comment parsing test cases in the testing suite?

01:09 <InPhase> teepee: Or is that basically impossible as it's unused from command line?

01:11 <InPhase> teepee: I was engaged in the fool's errand of cleaning up the last warnings from the build process, and the final one is this yacc shift/reduce warning from comment_parser.y, but it seems easy to mess up (in part from me not being skilled at .y) and I don't know what the full correct behavior is supposed to be.

05:49 <peepsalot> InPhase: iirc that warning is because there's an ambiguity between a slider range (with no step) [X:Y] , and a labeled drop-down with a single item

09:29 <peepsalot> InPhase: btw i blame thingiverse for the original customizer spec not requiring quotes on dropdown strings

11:48 <InPhase> peepsalot: That was 1 of the 3 warnings. The other 2 appear to be the part of the comment grammar that unrolls adjacent WORD and NUM elements in a manner that is ambiguous.

11:49 <InPhase> Unfortunately I don't understand this part's semantic goals, which is why I was looking for a test suite. It's easy to disable by declaring WORD and NUM events nonassoc, restricting their adjacency. But I don't know what that breaks, because I don't know what it was for in the first place.

11:54 <InPhase> I struggled last night to establish a precedence for the [X:Y] parse last night, but wasn't successful in large part because I don't understand what I'm doing, and the bison rules for establishing a precedence at the exact right spot appear to be a mysterious dark art. It is one of those problems which appears to be very common for people to have with their grammars, but for which zero of the

11:54 <InPhase> discussion threads leading to a resolution are sufficiently helpful.

12:02 <InPhase> I recall having some course content on this many many years ago, but alas, we did not cover tool use. And bison has a hefty processing delta between the input file and the state machines at which it reports errors, leaving it hard to discern where the ambiguity actually arises in the input. I tossed precedence markers at my top dozen or so guesses, but this did not seem to help.

12:05 <InPhase> I may have to just PR the other warnings away if I can't validate correctness of the comment parsing changes. My %nonassoc didn't break anything I could think of testing, but again, I'm not sure what that was for.

17:29 <gbruno> [github] t-paul pushed 1 modifications (Ignore error caused by expired Let's Encrypt root certificate.). https://github.com/openscad/openscad/commit/dee7a595316cfc7af52adfe4f4d3fce52cad1506

19:21 <gbruno> [github] t-paul pushed 1 modifications (Merge pull request #3909 from openscad/build-fix-macos