IRC logs for #baserock for Saturday, 2016-05-07

*** gtristan has quit IRC01:17
*** gtristan has joined #baserock01:53
*** locallycompact has joined #baserock02:11
*** gtristan has quit IRC02:43
*** gtristan has joined #baserock03:12
paulsherwoodi'm sliding towards box, bag, or sack of chunks08:14
locallycompactThe thing is08:52
locallycompactWe can get away with as many metaphors as you want08:52
locallycompactAs long as when you say sack of chunks you say why a sack of chunks is not a list of chunks or a set of chunks08:53
locallycompactAnd keep to the types08:53
paulsherwoodagreed. can we go with sack, then, for now?09:05
paulsherwood(since it's not too overloaded anywhere else... and the type enforcements can be done in the code)09:05
paulsherwood(would be easy to s/sack/new-name/ later if needed...)09:06
locallycompactmy point is that we need to if we can demystify the ontology for the user the types falls out naturally09:11
locallycompact*fall09:12
locallycompactso you call it a sack09:12
locallycompactif that's just a set or a list of chunks, then that's just a type synonym09:12
locallycompactthe user is still asking the same question just in a different interpretation09:14
locallycompact"is sack just a list?"09:14
locallycompactno it's kind of like a list but also this this and this09:15
paulsherwoodi think we are looking at different parts of this problem, still.09:17
paulsherwoodi want the code to be grokkable... so having simple names is easier for that. you want that the actual rules/logic for the names is locked down in code. i'm fine with that09:18
paulsherwoodeverything you want can be achieved irrespective of the choice of names/metaphors.... i'm just trying to avoid a repeat of our unweildy names/metaphors errors09:20
paulsherwoodi think sack fits the bill... and woudl be happy to help establish the rules/code for what a sack actually is in this context :-)09:21
locallycompactright, right09:26
locallycompactThe code should need only be derivative from the ontology is my point.09:30
paulsherwoodonce more in plain english, please? :-)09:34
paulsherwooddo mean that you want to ensure that the way we define 'sack' (ontology) should make the code that operates on sacks shorter, more obvious and more reliable?09:37
locallycompactthere should be a direct correlation between the intuition for a sack, the user description of a sack, and the type representation of a sack09:42
locallycompactthat's a deeper correlation than you might realise09:44
locallycompactthe intuition <-> type is guaranteed by the curry-howard isomorphism, nominally09:47
locallycompactThis is so that somebody who uses the tool won't have any problems understanding the logic, providing they speak rust's grammar.09:51
locallycompactIf they don't, they're just a user, but the intuition should still hold for the sack09:51
locallycompactIn the other direction, you can't just keep adding things to a sack beyond the point where it starts to feel like a sack because it would violate your intuition of a sack, then it's time to pick a new word/metaphor09:56
paulsherwoodok. but we can start with sack? i don't think creating a totally new word helps us much10:14
paulsherwood(and iiuc, the only way to satisfy your argument 100% would be to create a new word)10:15
locallycompactno no I wasn't suggesting that at all10:15
paulsherwoodok :)10:15
paulsherwoodi notice rust folks have settled on cargo and crates10:16
* paulsherwood wonders how well 'crate' in rust's lexicon maps to 'crate' in the rest of the world10:17
rjekA wooden box for storage and shipping?10:42
locallycompactA thing with a label on it that I dunno whats in it really but I could unpack it and stuff might fall out10:44
locallycompactACME10:44
rjekThat sounds like a box, in normal programming language terms10:45
locallycompactbox to me is just the name for the not quite right way to intuit monads and I don't really use it for anything else10:51
locallycompactI don't think10:51
locallycompactexcept cardboard box10:51
* locallycompact pushes some more wrongness to https://github.com/locallycompact/rust-ybd11:10
*** persia has quit IRC11:25
* paulsherwood notices that morph still uses linux-user-chroot, and hence fails to build devel on aws... 'too many mounts'11:32
*** persia has joined #baserock11:32
*** locallycompact has quit IRC13:03
*** locallycompact has joined #baserock13:04
*** gtristan has quit IRC15:28
*** gtristan has joined #baserock15:50
*** gtristan has quit IRC16:18
*** gtristan has joined #baserock16:18
*** rdale has joined #baserock18:17
*** locallycompact has quit IRC18:33
*** rdale has quit IRC18:49
*** gtristan has quit IRC20:20

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!