2017-01-10 UTC
# 00:23 Loqi aaronpk has 24 karma in this channel (1174 overall)
# 00:50 Loqi kevinmarks has 2 karma in this channel (261 overall)
KevinMarks, gRegorLove, chrisaldrich, KevinMarks_, tantek, loicm, cweiske, pfefferle, GWG, ben_thatmustbeme, Zegnat, dkm, arush and kline joined the channel
chrisaldrich joined the channel
tantek, mabkenar and arush joined the channel
# 13:33 rhiaro !tell aaronpk - do you have a way of automatically picking the best zoom level for a map to fit a path on?
# 13:33 Loqi Ok, I'll tell them that when I see them next
arush joined the channel
tantek and sebsel joined the channel
marcthiele and singpolyma joined the channel
# 15:07 Loqi aaronpk: rhiaro left you a message 1 hour, 34 minutes ago: - do you have a way of automatically picking the best zoom level for a map to fit a path on?
# 15:15 aaronpk start at max zoom level. subtract 1. check if everything on the map is within the window. if so, do it again.
# 15:16 aaronpk go until you eventually start clipping stuff off of the map
# 15:16 aaronpk also i'm checking based on the pixel locations, after having converted latlng to x,y pixels
# 15:16 aaronpk and you only need a bounding box, not the whole data, so the math is really simple
# 15:17 aaronpk now that i think about it, it might make more sense to do that the other way around. start with the smallest zoom, see if it fits. if not, increase 1 and check again
# 15:40 Loqi marcthiele has 1 karma in this channel (5 overall)
# 15:51 rhiaro it's taking LA to Japan the long way around the world though :)
# 15:53 rhiaro I dunno. Does the order of the coords in the path list make a difference?
# 15:55 aaronpk oh i see what you mean. no it just always has the -180 line on the left
# 15:58 petermolnar rhiaro, just out of curiousity, which hashing algo are you using to produce the "57fdb920c75c1" like urls?
tantek joined the channel
# 16:06 aaronpk i didn't realize you were going to just hotlink atlas maps :)
# 16:17 Loqi hotlinkingforthetimebeing has 1 karma
gRegorLove joined the channel
# 17:23 tantek like something that can be run as part of a commit hook or Travis or whatever for GitHub commits?
# 17:23 aaronpk an h-feed and h-entry validator would be super useful
# 17:26 tantek hopefully that makes that a bit more discoverable
# 17:27 KevinMarks Basically, he lost the h-entry on posts, so everything went into the feed
# 17:37 tantek ironically sounds like a similar level of problem that WordPress has/had
# 17:37 tantek with h-entry markup being not at the same level as the stuff inside
# 17:59 KevinMarks Not sure the indiewebify.me validator would work as an integration test
# 18:02 tantek KevinMarks, could you file an issue on indiewebify.me's github repo as to the problems you see with that, with suggested improvements if any?
# 18:12 KevinMarks I think the indiewebify.me approach is good for humans. I'd like advice from bear on good practice for a ci test validator
# 18:12 aaronpk it would be neat to have a validator that is intended to be used with integration tests. i don't know what that would look like tho
# 18:13 tantek presumably the same "backend" could be used for both
# 18:26 KevinMarks For ci you usually have known data to check against, whereas a validator is checking structure
# 18:29 aaronpk i could see doing something like "here's a URL. check whether you can find the "name" of this post and that it matches X"
# 18:32 bear I had the beginnings of a thought about that last month
# 18:32 bear using a test domain that was mostly staticly generated but with the python flask elements to build test domains for people
tantek joined the channel
# 18:49 KevinMarks I see a validator as a diagnostic tool. Protecting mastodon against the mf2 being accidentally refactored away is trickier.
# 18:50 bear agree with the need for both a validator and a test suite
# 18:50 bear they are two sides to the stable-spec coin
# 18:52 bear that's why having a static site works well for the CI side -- it's something that can be generated using various tools / libs and then given to a CI suite to test against
# 18:52 bear use text source files -> generate -> store -> standup with flask (or whatevs) -> run client side tests against temp site
# 18:53 bear both sides of the above can be stored as artifacts for others to use against their tools or generators
marcthiele joined the channel
# 19:24 KevinMarks That reminds me that I should take another look at the mf2 test suite
miklb joined the channel
# 19:38 miklb weirdest thing. Two domains on my new server, using same nginx conf file save the domain, and same certbot command also swapping domain and getting error on one of them (the one I really want to issue the cert for)
# 19:41 aaronpk are you sure there isn't some other nginx config block that's stealing that domain?
# 19:43 miklb hmm. Let me check if another nginx config
# 19:47 miklb usually is the obvious, but after seeing same error over and over, start looking for the difficult
# 20:01 tantek didn't barnabywalters have a bunch of utilities like that?
# 20:02 tantek extensions to php-mf2 that would return a twitter page as if it had microformats and was parsed accordingly?
# 20:02 tantek when you're adding this one-off support, are you adding it as fallback after no mf2 is found?
# 20:03 tantek such that if those sites started supporting mf2 explicitly, then that would be returned instead?
# 20:03 aaronpk if twitter suddenly added mf2 support i think we'd know ahead of time
# 20:03 tantek you could make that check after you find no h-* root objects
# 20:04 aaronpk also the twitter one is going to require API keys so it's not even up to XRay to decide that
# 20:04 tantek really? I thought barnabywalters's one didn't require API keys
KevinMarks joined the channel
# 20:05 tantek also for Twitter why not just do a URL substitution from the tweet permalink to the tweet intent permalink?
# 20:05 tantek and if you didn't know about that, I'm going to question your assertion "if twitter suddenly added mf2 support i think we'd know ahead of time" :)
# 20:06 aaronpk i do remember that :) I'd have to check that it actually includes all the parts of the tweet which i suspect it does not
# 20:15 tantek and if not, that's definitely a bug we can file with benward
Double_A, KevinMarks_ and marcthiele joined the channel
tantek joined the channel
# 23:01 aaronpk and google appengine doesn't run php 5.6 :headdesk:
# 23:05 petermolnar I digged out awstats and I kinda surprised it alive and maintained
# 23:06 aaronpk ah it's just that the default encoding was not utf-8 in php 5.5
KevinMarks joined the channel
# 23:28 aaronpk but it was just a charset issue. once i told it explicitly to use utf-8 then it worked
# 23:28 aaronpk tho i should probably test it on some fancy emoji. i wonder if twitter's character counters work with that
arush joined the channel