Ah, yes, ++can is better.

Or actually ++rep, if they're all just lsh 1?

I'm also somewhat confused what the surrounding code is doing: is render-hex-bytes literally just uncurried x-co:co? In that case, it's taking a list of @uxI(uint256), printing each as a tape of hex characters, concatenating the tapes, converting the concatenated into a cord of hex characters, splitting the cord into chunks 64 chars long, and then parsing the chunks back from hex into a tuple of the exact same numbers??

Yeah it looks like every single call to decode-topics accepts ~[%uint], ~[%uint %uint], or ~[%uint %address] where %address is an alias for %uint. While I'm all for future compatibility, I feel like in this case it would be reasonable to remove the function in question entirely and replace it with a series of length checks: ?> ?=([who=@ wer=address ~] topics.log) and similar.

(I'm also not sure why azimuth-events is a core instead of a (map @uxI term) which you could apply to event.log once and then idiomatically ?-. Or well, not literally term, %- my :~ result so that you can actually catch typoed(as currently) or unhandled(but known) event types)

…why is the compiler allowing lone terms in a $% o.o https://github.com/urbit/arvo/blob/master/sys/zuse.hoon#L7608-L7617
(I think the rune you technically want here is $?, possibly with a $@ split to $% tho tbh giving the commented static/dynamic precedence makes sense)

I completely forgot about rep. Serious tunnel-vision going on here.

Thanks for the feedback @ohAitch! I have touched up the logic here to not do superfluous @t <--> (list @) conversions.

Found an issue in the encoding logic while I was trying to test my changes there. We probably want more thorough tests for encoding, and tests at all for decoding. Official test data (which the current tests are sourced from) don't reach further than simple cases. Currently have a Q out to Ethereum core devs for recommended test data, but if that comes up blank, I might just pull in these.

Tests aside, this PR is ready for re-review.
(Edit: Actually ready now.)

I keep forgetting to write additional tests for this. I'll put that on my task-list, but merge for now, to at least get the fix in.