01:13 <simpson> Huh. So, when Monte's Capn parser loads a message, it builds an in-memory skeleton representing the basic pointer information. TIL that this is a dovetail with work done in the succinct data-structure community to speed up parsing.

01:16 <simpson> I'm watching ekmett talk about this sort of parsing: https://www.youtube.com/watch?v=uA0Z7_4J7u8

03:19 <kentonv> simpson, does that mean the parser traverses the whole tree upfront?

03:20 <simpson> kentonv: Mine only looks at the first header to determine where it should imagine its data and pointers to be. But after that, each additional bit of exploration generates a new bit of the skeleton.

03:21 <simpson> ekmett talks about the same idea, generating a mini-index on some JSON bytes lazily with Haskell's laziness.

03:21 <kentonv> ah

14:43 <nayrangnu> So, I was having a problem with an EtherCalc spreadsheet. The log reads:

14:43 <nayrangnu> node app.js --cors --vm

14:43 <nayrangnu> Falling back to vm.CreateContext backend

14:43 <nayrangnu> Please connect to: http://sandbox:33411/

14:43 <nayrangnu> Cross-Origin Resource Sharing (CORS) enabled.

14:43 <nayrangnu> Express server listening on port 33411 in development mode

16:55 <TimMc> Oof, kiled by the flood detector. Hope they come back...