#social 2017-10-29
2017-10-29 UTC
rowan and JanKusanagi joined the channel
xmpp-social, timbl, jankusanagi_, JanKusanagi and rowan joined the channel
rowan_ and rowan joined the channel
# puckipedia the triple store is backfiring
# xmpp-social [ajordan] yikes
rowan joined the channel
# puckipedia I fixed it by requiring the triple store and database behind it are always in sync
# puckipedia now I need to fix a few things that are a limitation of the ORM I based the thing on
# puckipedia yep this was so much a bad idea
# puckipedia I think I'm going to throw away the whole idea of "committing changes" and just use transactions everywhere
# puckipedia works too
# puckipedia haha
# puckipedia I am actually thinking of redoing the entire paging system
rowan and bengo joined the channel
# xmpp-social [ajordan] I feel like I'm watching a transcription of my entire life
# xmpp-social [ajordan] "I swear to used to work"
# xmpp-social [ajordan] "/me shakes fist at past me"
# xmpp-social [ajordan] too real
# xmpp-social [ajordan] I sympathize rhiaro :P
rowan_ joined the channel
# puckipedia hm, only 17 seconds to POST an object to my outbox
# puckipedia 10 seconds now
# puckipedia good enough(TM)
# puckipedia yyyyep
# puckipedia I'm going to have to spend like a million hours figuring out how to get a triple store down to <1 second for a request
# puckipedia yepp
# puckipedia yep. postgres, currently
# puckipedia yeah, I've actually built it on top of EntityFramework, a .NET ORM
# puckipedia I almost directly move from the quad store into JSON-LD
# xmpp-social [ajordan] puckipedia: how good are .NET's performance profiling tools?
# puckipedia ajordan: no idea(TM)
# xmpp-social [ajordan] Classic :D
# xmpp-social [ajordan] I mean same though
# puckipedia I have three tables, first is a table
{ int id, string uri }
for all the ID values# puckipedia (e.g. type, object id, parameter id)
rowan joined the channel
# puckipedia aha. I think just adding an index on the Uri, and adding an index on SubjectEntityId should help a ton
# puckipedia SELECT "a"."AttributeId", "a"."Uri" FROM "Attributes" AS "a" WHERE 0 = 1
# puckipedia except I still don't know what this is doing
jankusanagi_ joined the channel
# puckipedia ohaio!
# xmpp-social [ajordan] Heya cwebber2!
# xmpp-social [ajordan] :-)
# xmpp-social [ajordan] Happy to hear your weekend was good \o/
# cwebber2 gosh I'd really love to close https://github.com/w3c/activitypub/issues/260
# puckipedia yeah. I already have this loose relationship between Actor and user :P
# xmpp-social [ajordan] It seems like they don't quite understand some of the process stuff going on
# xmpp-social [ajordan] Like the fact that we're already in CR
# xmpp-social [ajordan] And also it seems like they want migration to be built in to the data model when in reality that's not *really* possible due to the coupling of URLs to ids
# xmpp-social [ajordan] So we do it "imperatively" instead
# xmpp-social [ajordan] Is that just me?
# xmpp-social [ajordan] Right
# puckipedia firefox's inspector broke
# puckipedia anyways. I think I got a request down from 4s to 3s now
# xmpp-social [ajordan] But at least atm the spec is, in practice, all http
# xmpp-social [ajordan] Right sorry
# xmpp-social [ajordan] I meant that all current implementations are https://
# xmpp-social [ajordan] And will be for a while
# xmpp-social [ajordan] So practically speaking we can't do anything that would put migration in the data layer into the core spec because it would break the real world
# xmpp-social [ajordan] That's what I meant
# puckipedia me too
# xmpp-social [ajordan] Hah
# xmpp-social [ajordan] https://twitter.com/sadserver/status/672846409444364288
# Loqi [@sadserver] Happy Friday! Share a gif that shows how your last deploy went https://pbs.twimg.com/tweet_video_thumb/CVZuPk9UAAEQICx.png
# xmpp-social [ajordan] ^^^ us rolling out a new CR :P
rowan joined the channel