#dev 2018-03-24

2018-03-24 UTC
#
snarfed nope, that's exactly why we locked github backfeed down to require synd links early on
#
tantek ah right, already solved by that
#
tantek after the adactio hilarity
#
tantek I still wonder if he ever noticed
leg joined the channel
#
tantek snarfed, hmm just had a publish failure. investigating my side of it ...
[mrkrndvs] joined the channel
#
tantek odd - looks like https://brid.gy/github/tantek says I need to "Click to enable" both listening and publishing
#
Loqi tantek
#
tantek I wonder what I did to disable that
#
snarfed ah yeah, your tokens somehow didn't have the notifs scope, which listen needs
#
tantek clicked and didn't need to re-auth
#
tantek retrying publish...
#
snarfed not sure how, maybe beta quality iteration hiccoughs
#
snarfed sorry!
#
tantek no problem at all - I need to get better about checking the specific error returned to see if there's a way to auto-redirect to re-auth
#
tantek er, re-enable
#
tantek or I wonder if there's a way to preflight that
#
tantek is there a way to ask Bridgy Publish if I *can* publish without actually doing the publish?
#
tantek bridgy++ for handling labels with spaces :) https://github.com/microformats/h-entry/issues/7
#
Loqi bridgy has 5 karma in this channel (69 overall)
#
Loqi [tantek] #7 Can we improve h-entry backcompat to handle common WordPress hentry streams?
#
tantek hey toomim didn't mean to slow your momentum :)
#
tantek aaronpk did you mean http://www.manton.org/ as an example?
#
tantek yes just url and content
#
tantek the articles also have name
#
tantek and the dt-published appears to be unstructured inside rel=bookmark
#
tantek or rather, a chunk of ISO8601-ish inside
#
tantek so we'd have to handle doing something like ignoring non-numbers, then parse for ISO8601 date time with slightly more flexibility (allow "/" as y-m-d separators instead of "-", look for "pm" or " pm" immediately after time string and handle per vcp date-time parsing
#
tantek also a bit surprised that's in the 'entry-content' instead of the 'entry-meta' - I wonder why manton chose to do that - because I think normally themes put permalinks / dates in the entry-meta
eli_oat joined the channel
#
tantek hmm - his article post has several rel-tag links which are not being picked up by the parser for some reason
#
tantek and the article also has time.entry-date[datetime] for dt-published which is already in the proposed list for backcompat
#
tantek ah the article also has <span class="author vcard"> which is getting picked up by the parser
#
tantek !tell manton any particular reason your notes put permalink/published date inside the entry-content instead of the entry-meta like your articles do on http://www.manton.org/ ? (also your articles have author in entry-meta, whereas your notes don't seem to, deliberate? or theme omission?)
#
Loqi Ok, I'll tell them that when I see them next
snarfed joined the channel
#
snarfed tantek: if your account is disabled, the bridgy publish wm will return 400 with body starting with "Publish is not enabled for your account."
#
snarfed and yeah you can ask it preemptively. https://brid.gy/about#api , p-bridgy-publish-status
#
tantek awesome ok I will capture that as a something to publishing pre-check
#
tantek.com edited /silo-quits (+542) "SpaceX and Tesla Facebook quits" (view diff)
[snarfed] joined the channel
#
[snarfed] tantek: maybe low priority. i think it was just an early bridgy bug, i don't expect it regularly with the github API
#
[snarfed] unlike eg Facebook which you have to reauth every 2 mos
#
tantek yeah I'd generalize it to pre-publish check anything that was going to auto-POSSE via Bridgy Publish
#
tantek better that the user be given a heads-up so they can have a chance to update the auth if needed before pressing Publish
#
tantek especially in the case where multiple POSSE destinations are involved and it would be annoying to have only some fail due to expired auth
#
tantek (e.g. for me, photos)
#
loqi.me edited /silo-quits (+161) "tantek added "https://twitter.com/elonmusk/status/977211923719598086" to "See Also"" (view diff)
#
loqi.me edited /silo-quits (+124) "tantek added "https://twitter.com/RMac18/status/977217618649137152" to "See Also"" (view diff)
#
loqi.me edited /silo-quits (+124) "tantek added "https://twitter.com/RMac18/status/977217884526067712" to "See Also"" (view diff)
#
tantek.com edited /silo-quits (+329) "spacex/tesla related tweets and screenshots" (view diff)
#
loqi.me edited /algorithmic_feed (+419) "tantek added "2018-03-22 The Verge: [https://www.theverge.com/2018/3/22/17152660/facebook-politics-artificial-intelligence-algorithms If Facebook controls your mind, so do a lot of other tech companies] <blockquote>… François Chollet, an artificial..." (view diff)
#
tantek.com edited /algorithmic_feed (+71) "rest of see also quot" (view diff)
[kevinmarks] joined the channel
#
tantek.com edited /TLS (+304) "stub, for jargon" (view diff)
#
tantek.com moved /TLS to /Transport_Layer_Security "prefer expansion of abbr"
#
tantek.com edited /ssl (+19) "update redirect since jargon" (view diff)
#
tantek.com created /certificate (+36) "r" (view diff)
#
tantek.com created /cert (+36) "r" (view diff)
#
tantek.com created /public_key_certificate (+371) "stub jargon" (view diff)
renem_ joined the channel
#
tantek since SSL is officially *prohibited*, should we even use the term informally when referring to e.g. ssl certs? or should we instead refer always to TLS, e.g. TLS cert? cc: snarfed bear
#
tantek (I'm looking at all the uses of "SSL" as a word/adjective in /HTTPS and wondering if those should be changed/removed)
snarfed joined the channel
#
aaronpk I wasn't thinking of any in particular, I saw that problem in a whole bunch of wordpress sites, not just Manton.org
tantek, snarfed, [jeremycherfas] and [unoabraham] joined the channel
#
[unoabraham] If you mean IndieAuth, then yes. The domain is not the identity, it takes it from either Twitter or Github, haven't faced any issues myself.
[kevinmarks] joined the channel
#
[kevinmarks] Did we add the h-feed backcompat suggestions?
#
[kevinmarks] Right we did.
#
[kevinmarks] With WordPress you could parse a date out of the post url, though that is a bit gnarly
#
[kevinmarks] That was one of my technorati fallbacks
tantek and [unoabraham] joined the channel
#
Zegnat Anyone happen to have a 1 file PHP script for POSSE to Twitter at their fingertips? Before I start searching...
#
sebsel With or without Bridgy publish?
#
sebsel If Bridgy publish, it's just adding mf2 and sending a webmention.
#
Zegnat Without external services.
#
sebsel Not me, then :)
#
Zegnat Yeah, I was thinking about that. Might end up using it for iteration one.
AngeloGladding, eli_oat, leg and tantek joined the channel
#
aaronpk [unoabraham]: its the other way around. The domain is the identity and uses twitter or GitHub to authenticate that identity
#
aaronpk This is demonstrated by the fact that you can log in to the wiki with the same user account by authenticating with either twitter or github but you end up logged in to the same account either way
[kevinmarks], snarfed, [kim_landwehr] and eli_oat joined the channel
#
@Audiopedia_ What is WEBMENTION? What does WEBMENTION mean? WEBMENTION meaning, definition &amp; explanation: http://youtu.be/9CsswZ2p0Wk?a via @YouTube (twitter.com/_/status/977575923686916096)
eli_oat, snarfed, [kevinmarks], KartikPrabhu, AngeloGladding, deltab and tbbrown joined the channel
#
bear.im edited /public_key_certificate (+159) "expand on the definition of TLS and when its used" (view diff)
#
bear.im edited /Transport_Layer_Security (+141) "add a bit more to the definition and use cases" (view diff)
#
bear yes - anyplace you used to use SSL should be changed to be TLS (it's more accurate and enforces the fact that SSL is now deprecated)
#
aaronpk it's going to take me a long time to un-learn saying "SSL cert"
#
sknebel I mean it's still a cert you can use for SSL. You shouldn't, but the cert is the same ;)
#
bear :)
#
aaronpk heh
leg, eli_oat, gRegorLove_, jonnybarnes and tomasparks joined the channel
#
GWG Who is posting audio right now?
[jeremycherfas] joined the channel
#
[jeremycherfas] Me. Well, not right now, but once every two weeks sometimes more frequently.
#
GWG I want to improve my presentation
#
[jeremycherfas] I currently rely on the player built in with the Blubrry plugin, but I am not super happy with it. I have not had any complaints. I also have no idea how many people listen there rather than via a podcatcher.
#
KartikPrabhu <audio> element should give a default player
#
[jeremycherfas] Yes, and requires more fiddling.
#
GWG [jeremycherfas]: I've used that too, but I want to do my own display.
[snarfed] and [kevinmarks] joined the channel
#
Zegnat Anyone want to tell me why my regex is not working?
#
Zegnat I am doing a replace with /( +(?=\n)|(?<=\n) +|^[\t\n\f\r ]+|[\t\n\f\r ]+$| +(?= ))/ but it isn’t removing a linebreak at the start of a string. While it is supposed to do so...
#
sknebel hm, looks like it should do that
#
sknebel although I seriously wonder what you are doing ;)
#
Zegnat Yes, but I am failing my test
#
Zegnat https://wiki.zegnat.net/media/textparsing.html - step 2, 3, and 4 of “Plain text of element” in a single replace.
#
Zegnat Obviously ;)
#
AngeloGladding lookbehinds aren't supported in js?
#
Zegnat PHP
#
AngeloGladding https://www.regexpal.com/
#
sknebel well, in the regex101 tester it does remove them
#
AngeloGladding ah
#
Zegnat Yeah, I tried there too sknebel. So I have no idea why it isn’t doing what it is supposed to do when I run my tests
#
sknebel code?
#
Zegnat https://gist.github.com/Zegnat/bd1c53df51568d3b5a9c0a34c2c2b643
#
sknebel also, 2,3,4 there says nothing about removing a leading LF?
#
Zegnat Oversight on my part
#
Zegnat Let me update my updated textparsing.html
#
Zegnat I changed step 3 to “Strip leading and trailing ASCII whitespace from output.”
#
Zegnat Which is actually what would be expected. No odd errant newlines
#
Zegnat Which should be covered by ^[\t\n\f\r ]+|[\t\n\f\r ]+$
#
AngeloGladding you can't pass it through a generic strip function first?
[cleverdevil] joined the channel
#
Zegnat I could. But where is the fun in that when you have to use regex anyway? ;)
#
sknebel readability?
#
Zegnat I actually probably will switch to a trim()
#
Zegnat But I now want to figure out why this regex isn’t working in the first place
#
AngeloGladding if it breaks now it'll break when you least expect it :)
#
AngeloGladding but i understand the itch
#
Zegnat But it shouldn’t be breaking at all. It honestly is a pretty straight forward regex
#
Zegnat Hmm. I wonder if the order is important for PHP, when the regex matcher walks through the string
#
Zegnat That’s it!
#
Zegnat TIL
[gerwitz] joined the channel
#
[gerwitz] Is there a consensus on what error responses should be used if a Micropub endpoint receives a valid request that it doesn’t support?
[jjdelc] joined the channel
#
[jjdelc] sounds like 406?
#
sknebel no, 406 is *only* for Accept-header mismatches
snarfed joined the channel
#
sknebel I'd just do 400 with an appropriate message
#
sknebel if there is no way to accept the request - what specific case are you thinking of?
#
aaronpk https://www.w3.org/TR/micropub/#error-response
#
aaronpk basically HTTP 400 for everything that isn't authentication errors
theReal33mhz joined the channel
#
theReal33mhz Pnut.io now has an official third-party web client that properly uses microformats for posts and profiles, e.g., https://beta.pnut.io/@shawn/posts/311679 and https://beta.pnut.io/@33mhz
#
Loqi [Shawn Throop] Blasting Skrillex’s Bangarang to cover up Nutcracker music, in March. I have an odd profession…
#
snarfed theReal33MHz++ cool!
#
Loqi thereal33mhz has 1 karma in this channel (2 overall)
#
theReal33mhz The karma is real! X)
leg, leg1, [kevinmarks] and [kimberlyhirsh] joined the channel
#
snarfed just released a new version of granary with python 3 support! https://pypi.python.org/pypi/granary (cc [cleverdevil])
#
snarfed took way more work than it should have. glad it's done.
#
sknebel woah
#
sknebel snarfed++
#
Loqi snarfed has 52 karma in this channel (371 overall)
#
snarfed oh btw aaronpk re watchtower polling tiers, i realized you maybe don't need to keep a bunch of old post ids per feed. you could keep just the latest one, use a regexp (i know i know :P) per format to find the latest id in a feed, and check that
#
snarfed just another alternative
[miklb], eli_oat and [snarfed] joined the channel