2016-04-01 UTC
tantek, danlyke_ and snarfed joined the channel
# 01:00 tantek see salmention discussion earlier today in the logs
# 01:01 tantek there's actually still room for experimenting with salmention sending in different cases (e.g. up/downstream to previous likers) specifically because people see similar notifications!
# 01:01 tantek E.g. someone commented on a post you are following.
# 01:02 tantek where "following" a post can happen a few different ways, including by liking it
# 01:02 tantek or by explicitly clicking the little v down arrow in the top right of a post and asking to get notifications (or stop getting)
# 01:03 tantek lol at the transform the -of suffix discussion ;)
# 01:09 tantek shaners++ good work on getting something up for IWC LA 2016!
# 01:17 tantek aaronpk, snarfed re: salmentions - your discussion is making me realize that I need to be even more proactive about documenting all the different kinds of notifications that various silos do like FB, IG, Twitter
# 01:17 snarfed tantek: oh boy that's a big hairy ambitious moving target
# 01:17 tantek because frankly, most of those are precisely why we have salmentions (including from SWAT0 scenarios)
# 01:17 tantek snarfed: it's not that bad, they don't add too many too often
# 01:17 snarfed definitely useful! but maybe don't stress too much about being comprehensive
# 01:18 tantek plus I'll obviously start with the more common ones
# 01:18 tantek or very interesting / odd ones - like the location suggestion stuff!
# 01:18 tantek anyway we need this empirical data to shape our protocols like salmention, instead of guessing at what ifs and what shoulds
# 01:19 tantek KevinMarks: we can start with only the Twitter notifications we like :) (that is, that we'd like our sites to do for us)
# 01:19 Loqi tantek meant to say: (instead of all their growth hacking garbage)
# 01:35 tantek whoa I just figured out that Facebook uses "reply" as the generic term!
# 01:36 tantek as in when I RSVPd as "interested" to an event, and a friend commented on it
# 01:36 tantek the notification I got / saw was: "so-and-so commented on your reply in Event-Name"
# 01:37 tantek no kevinmarks you were saying something else - that "reply" is more specific! as in a reply to a comment
# 01:37 tantek thus implying that reply is a superset of things including RSVPs
# 01:37 tantek also that whole theory about comment being 1st level, and reply being 2nd level is also shot
# 01:38 tantek because "commented on your reply" is implying that comment is 1 more level than the reply!
# 01:38 tantek gotta make dinner, will upload screenshots later
# 01:39 tantek see above ^^^ for some recent screenshots / documentations of various FB features
# 01:39 KevinMarks "commented on your comment" and "replied to your reply" sound odd
# 01:39 tantek hah KevinMarks that's funny. Sure, I'll look for "replied to your reply" type notifications as well ;)
emmak_ joined the channel
mblaney and snarfed joined the channel
# 03:17 acegiak snarfed. feel free to submit PRs for things like reacji on sematic linkbacks
# 03:18 acegiak I'm having the realisation that I have too many projects considering I start my new job on monday
# 03:46 GWG I don't think I am as concerned about reacji
mblaney1 joined the channel
snarfed joined the channel
# 04:54 Loqi The following is a guest post by Pankaj Parashar. Pankaj wrote to me about some pretty cool styled progress elements he created. I asked if ...
# 04:54 snarfed KevinMarks: oh god yeah, sure, but that would need ajax calls to update it
# 04:56 snarfed KevinMarks: the dirty secret is that the whole conversion happens synchronously within the single http request
# 04:56 snarfed converting it to async is the big task that, just, sorry, nope
wolftune joined the channel
# 05:33 KevinMarks not with the app running as it is a giant dependency bundle; I tried the html+css
# 05:51 snarfed KevinMarks: fun! it mostly works. the scroll bar moves down the page due to some other parts i can probably fix. thanks for the nudge!
# 05:51 snarfed also the animated gradient is pretty slick; i'm guessing that's just firefox's own rendering of <progress>, new to me
# 05:53 KevinMarks you could do the same trick, or make the <div class="progress"> around each stage and change the selector from #progress to .progress
# 05:54 snarfed hmm, it was moving down with just blank space above it (beyond the first few lines of text)
# 05:55 snarfed anyway. i'll be able to look more tomorrow, but feel free to poke more in the meantime!
# 05:56 snarfed setting it up locally shouldn't be too bad, just `sudo pip install boto webob youtube-dl` and then point any web server at app.py as CGI
# 05:57 snarfed just replace huffduff-video.snarfed.org in the bookmarklet with localhost, and configure your local web server to point /get to app.py as CGI.
# 05:59 snarfed right, it will break before uploading to s3, but should be enough to start testing
# 06:00 snarfed if you get ambitious, you can change the constants to point to your own bucket and fill in the aws_* files with your keys. but definitely not necessary
mlncn_, jihaisse, cweiske and loic_m joined the channel
friedcell and KartikPrabhu joined the channel
lewisnyman, Pierre-O, M-kegan, adactio, loic_m and j12t joined the channel
adactio, John_Duh, Pierre-O, kerozene, friedcell, j12t, Kopfstein, Garbee, lewisnyman and camerongray joined the channel
tantek and j12t joined the channel
# 13:26 Loqi Got it! There are now 48 spammers blacklisted
snarfed and mblaney joined the channel
hs0ucy, lewisnyman, cphuntington97, gavinc, danlyke_, PFMurph, shiflett, wolftune, miklb and mlncn joined the channel
j12t joined the channel
# 15:20 aaronpk [shaners] just added the Pivotal logo to the footer and event page, thanks!
tantek, j12t and snarfed joined the channel
omarhuanca1 and omarhuanca joined the channel
# 16:10 omarhuanca I'm sorry, tantek
omarhuanca joined the channel
# 16:16 tantek omarhuanca: what's the specific publishing use-case you're trying to solve?
# 16:18 omarhuanca my project manager say me, add any microformat for lyrics with startTime, endTime.
# 16:18 omarhuanca it was understood
# 16:20 omarhuanca initialy, I try use hListing
# 16:23 omarhuanca for karaoke
# 16:23 tantek hmm - I don't think that's a good match, hListing is for listings in the sense of Craigslist
# 16:23 tantek (which I realize is yet another overloading of "list")
# 16:31 omarhuanca ok, talk more later.
snarfed joined the channel
# 17:03 snarfed eyes the bridgy beta testers on instagram scraping
# 17:03 snarfed happily sending wms, looks like everything is humming along
# 17:06 Loqi [Aaron Parecki] The #unipiper dawns a #onewheel #selfbalancing #hoverboard because #pdx
# 17:06 aaronpk also oops apparently my video posts aren't working for some reason
# 17:09 aaronpk that latest post was a video but my site only got the photo
# 17:10 snarfed looks like instagram's html only includes the latest 10 likes. not sure how those 10 are chosen/ordered. :/
# 17:11 snarfed so IG like backfeed may be incomplete just like API was :/
# 17:13 Loqi [davidjohnmead] tag:instagram.com,2013:1218518284778719564_1500881_liked_by_256750317
# 17:15 aaronpk tho I should make Loqi recognize when a post is a like/repost of a post and say something to that effect
# 17:17 aaronpk so that's actually an error from XRay on appengine
# 17:17 aaronpk from a googler: ""RPC Error: Unknown error - 4" is strange, RPC error 4 is returned when the response from the HTTP request exceeds 32MB but we should be returning "response to large" rather than "unknown error"."
# 17:18 snarfed is this the appengine-to-appengine php http request problem where you need to set a special header?
# 17:19 aaronpk possibly? but i'm confused why it would work sometimes but not others
# 17:20 aaronpk also what's frustrating about that is I have no way to know when a URL is an appengine URL
# 17:20 snarfed (technically in this case you do know because appspot.com, but in general, agreed)
# 17:21 aaronpk yeah when i was testing XRay I kept encountering URLs which are CNAMEd to appengine
snarfed, KartikPrabhu, mlncn, hs0ucy, wolftune and gRegorLove joined the channel
[shaners] joined the channel
# 18:42 KevinMarks I switched from using requests to using the fetch api directly, as I realised that on appengine requests was shimming down to it anyway
# 18:42 aaronpk i'm pretty upset at appengine network stuff in general
# 18:43 KevinMarks it's a PaaS vs VM thing; when you do get to piggyback on the crawl infrastructure and get freshly cached stuff it's good
# 18:44 aaronpk when everything works fine it's fine, but the error handling is awful
omarhuanca and mlncn joined the channel
# 18:47 aaronpk i have two different ways I make HTTP requests now, depending on whether i'm talking to an appengine URL
# 18:48 snarfed it's almost certainly php-specific; i've never had this problem in python or go
# 18:48 snarfed aaronpk: can you not just use the app engine specific headers etc for all urls? or are they bad for non-gae urls?
# 18:49 aaronpk the problem is i have to tell the http library to not follow redirects when fetching appengine URLs
# 18:49 snarfed such a weird bug. and weird that they've let it sit instead of handling it themselves.
# 18:50 aaronpk IIRC it would just return a weird error otherwise
# 18:55 KevinMarks may still be worth doing that as a fallback in case you guess wrong about an appengine site; does it fail fast?
# 18:55 aaronpk yeah. i'll probably switch to that and see how it works
Breadbasket_, barnabywalters and tantek joined the channel
begriffs joined the channel
# 19:43 tantek hmm I thought we had documented/proposed way of moving forward on the over-implied p-name problme
# 19:44 tantek I'm still looking for issue / proposal / discussion
# 19:45 tantek but that only really converged on u-url implying
# 19:55 tantek looks like the any explicit p-* = no implied p-name would stop the noise from Bridgy's markup
# 19:56 snarfed also maybe don't overfit to bridgy's markup, since bridgy mf2 pages aren't intended for people, and are rarely pasted here
# 19:57 tantek we're not trying to fit it to Bridgy, but rather it's just one of many data points
# 19:57 aaronpk also bridgy's markup is totally intended for people, since peoples' web pages parse the bridgy markup and then display comments and things
j12t joined the channel
# 19:57 snarfed aaronpk: sure but that's indirect. as opposed to eg posts or h-feeds that are rendered by browsers directly
# 19:57 tantek people (or tools/services) shouldn't have to explicitly include an empty p-name to avoid this problem
# 19:58 aaronpk bridgy's html markup is not intended for viewing, but the mf2 markup certainly is
# 19:59 tantek snarfed, pretty sure no current consumers of Bridgy depend on the legacy u-like markup so we should probably drop that (to avoid creating new consumers reading it!)
# 19:59 tantek e.g. s/class="u-like u-like-of"/class="u-like-of"
j12t joined the channel
# 20:12 tantek hmm, not sure how to do this fix in Bridgy so I'm going to leave that one to you snarfed
# 20:56 Loqi [gRegor Morrill] @newsgenius How do we block News Genius on our sites like @NYTimes does?
# 20:57 gRegorLove I also sent it to one of their community managers. Still silence.
# 20:57 tantek aaronpk: any way to get Loqi to regenerate this week before it goes live again?
# 20:59 tantek this looks a bit odd: "Saturday, April 16 - April 17"
# 21:06 gRegorLove KevinMarks: There's some other way to "block" News Genius that's not apparent to the public. Try a NYT URL with it. It doesn't redirect or anything, but shows the page doesn't support it.
# 21:06 gRegorLove So there's some semi-official way to block that News Genius is not telling people about.
# 21:07 tantek so it doesn't work on big news company sites that it was supposedly targeted at, but does work for people ganging up and harrassing small indieweb sites
# 21:07 tantek really they should have just called it Site Vandal
# 21:09 tantek can't tell if Loqi is waiting for more edits, or a new publication time or, maybe is out to lunch?
# 21:12 tantek Perhaps there's also an opportunity for a browser plugin that automatically removes the genius domain wrapping of any link that uses it
# 21:14 gRegorLove That doesn't help the authors, which is what I was looking for, more broadly.
# 21:20 aaronpk i'll regenerate the HTML for the archives but looks like the email went out at 2:03
# 21:24 tantek aaronpk: looks like u-url mis-scraping from the nest h-card u-url instead of parsing from the h-event u-url
# 21:24 KevinMarks u-url is on the bit outside the span, not the link in the p-name
# 21:26 tantek the first u-url is *on* a span of *just* the URL
# 21:29 tantek it doesn't need to have anything to do with the link in the p-name
KartikPrabhu joined the channel
# 22:20 tantek aaronpk: based on today's experience I think we could use an hour's warning
# 22:21 tantek alternative: post the preview at noon which gives people their (Pacific Timezone) lunch hour to madly edit it in a rush, then post another preview at 13:00
wolftune and trying joined the channel
# 23:43 Loqi [Aaron Parecki] Guess what day is the worst day to announce you're quitting your job