2018-03-22 UTC
leg, snarfed, eli_oat, [tantek], sebsel and tantek joined the channel
tantek joined the channel
tantek, renem, j12t and EmreSokullu joined the channel
snarfed joined the channel
[unoabraham], tantek, KartikPrabhu, cweiske, voxpelli, [pfefferle] and Kongaloosh joined the channel
dgold and jeremycherfas joined the channel
[kevinmarks] joined the channel
swentel joined the channel
# 09:53 cweiske dgold, I see you offer atom feeds. I use /stapibas to automatically send out webmentions to all links in my atom feeds. shall I add your feed, too?
# 09:55 cweiske jeremycherfas, do you offer a feed on your main page? I didn't find one
# 09:58 cweiske could you make the atom and rss links machine discoverable?
# 09:59 cweiske jeremycherfas, I could also add your blog's atom feed to my stapibas instance, so that webmentions get sent out automatically
# 10:01 cweiske but feed readers cannot discover that they link to the feeds
# 10:03 cweiske I know that all readers support head links. no idea about normal <a> with rels
# 10:13 cweiske file extension does not matter. it only matters that it is atom
# 10:14 Loqi cweiske has 27 karma in this channel (129 overall)
petermolnar joined the channel
# 10:34 cweiske and your atom feeds have <author><name>John Appleseed in them
# 10:35 jeremycherfas What the? That is weird. See, you fix one thing and another pops up. Its whack-a-mole.
# 10:35 jeremycherfas Book & should be Books &c -- that is, books and other reading related things.
# 10:36 cweiske the title is for users, and they should not care about the format
# 10:39 jeremycherfas I guess the Atom feed is picking up authorname from somewhere strange, and I think I know where.
# 10:39 cweiske dgold, I've added you. it will start pinging in a couple of minutes.
[kevinmarks] joined the channel
# 10:45 cweiske there is a bookmarklet that you can use to view the mention status of your posts
# 10:45 jeremycherfas About stapibas: when I put a url into Telegraph, it shows that it could send webmentions to things like tag pages and category pages. I never want to do that, so I don't send the webmention. Would stapibas send those automatically?
# 10:46 cweiske stapibas only looks at the links inside of h-entry
# 10:46 dgold I may have to look at my h-entry markup, but I _think_ i'm good
# 10:47 jeremycherfas I'd better check mine too. By extension, you are saying that Telegraph looks at everything for links.
# 10:48 Loqi [Daniel Goldsmith] I agree that this time feels different, but the truth is that as we have become more compartmentalised Facebook has become the touchstone of community, of belonging. People move to new places and Facebook is where they meet their neighbours. The effe...
# 10:49 cweiske stapibas ONLY pings links inside h-entry and then e-content
# 10:50 cweiske unless you put category links inside e-content, which you should not do.
# 10:51 jeremycherfas Those links I was talking about appear to be inside h-entry, in the "html" of "content"
# 10:56 Loqi cweiske has 28 karma in this channel (130 overall)
# 10:59 Loqi cweiske has 29 karma in this channel (131 overall)
# 11:00 dgold if its working inside that to try to ping colinwalker's site, then all is well :)
# 11:47 jeremycherfas Very interesting. Prodded by cweiske, I've had a much closer look at my Grav theme, and it is clear that I can make the mark-up much more logical. But that's also a bigger task than I am ready for today. Still, it creates a significant itch.
# 11:51 petermolnar perl 5 with moo/moose is surprisingly efficient, although still hm... interesting to read.
# 11:58 dgold just realised my aperture doesn't have the [Notifications] Channel, per the spec
# 12:00 cweiske dgold, there is a catch currently. you have to tell stapibas that it should update your feed manually.
# 12:02 cweiske I have a script on my server that triggers this when a new blog post is rendered
# 12:03 dgold micropub->nanopub->content entry; incron notices the new file has been written and 'released', this triggers a rebuild
# 12:03 cweiske yes, than this trigger should also do that curl call
# 12:03 dgold but the script for that can be changed to do a curl `ping`
barpthewire joined the channel
# 12:47 Zegnat I think I am getting the hang of this TDD. mf2 parsing problem? Write a failing test. Fix it. Commit.
[kevinmarks] joined the channel
# 13:02 [kevinmarks] now what I really want is common tests for all parsers, but the existing ones are a bit of a mess
# 13:04 Loqi [microformats] tests: Microformats test suite
# 13:07 cweiske I'm also planning to implement a mf2 parser, so having that test suite in a good state would be very helpful
leg joined the channel
eli_oat joined the channel
# 13:09 sknebel Zegnat: good start would be submitting all tests you add/propose for other parsers also to the central repo
# 13:11 sknebel (I assume you are not just doing it for "I have nothing else to do" ;))
# 13:12 Zegnat may or may not also have half an alternative PHP parser because of NIH
# 13:12 Zegnat Also because I didn’t want the xpath logic used by the current one, and instead wanted to see if it was possible in a single complete tree walk
# 13:14 Zegnat I’d say DOM style. That is, you give the parser a DOMDocument, and it walks it using only standard DOM methods to get the microformats out
# 13:15 Zegnat Ah, no, my microformats parser works on a DOM tree. Not event based off of a parser walking the raw HTML
# 13:34 Zegnat I wonder if I could make php-mf2 run through the centralised tests using PHPUnit
# 13:36 cweiske there'd be people in here giving you karma for that
# 13:43 Zegnat Oh, wait, we are on PHPUnit 4. I need even older documentation, possibly
# 13:43 Zegnat Alright, I’ll try with this documentation in hand first
# 14:19 Zegnat Lets see if that improves when I stop doing string comparisons of the resulting JSON and do an actual proper comparison of JSON objects.
# 14:20 sknebel With the recent changes theres going to be a bunch of actual defects
snarfed joined the channel
# 14:24 Zegnat That’s fine, as long as I know the failures aren’t caused by my checking code. Then I can start filing fixes to the centralised tests
# 14:26 skippy when one uses Quill to micropub a note with a photo, the photo is an element in the array the gets posted. How do folks reference said photos in the body of the posts they are composing?
# 14:28 skippy also, woah, just saw that quill supports post-by-email? that's slick.
leg joined the channel
# 14:44 Zegnat Almost halved the amount of failing tests already :D
# 14:44 Loqi testing has 7 karma in this channel (26 overall)
# 14:44 Zegnat I just have to be careful. Because I am currently fixing tests by tweaking how the tests run ...
# 14:46 Loqi spaces has -1 karma in this channel (2 overall)
# 14:46 Loqi testing has 8 karma in this channel (27 overall)
[manton] joined the channel
# 14:57 Loqi kartikprabhu has 15 karma in this channel (187 overall)
# 15:25 Loqi [Zegnat] #163 Run mf2/tests test suite with PHPUnit during default testing
# 15:30 Zegnat If the improved rel parsing PR gets merged in it actually fixes 1 of the test suite tests. Nice to have this working
# 15:34 sknebel At least that's what the commit addibg it sounds like
# 15:34 KartikPrabhu also the tests don't test for specific things so they are not very useful (atleast for me)
# 15:38 sknebel Yes, sounded like we have green light to change that tho
# 15:38 KartikPrabhu sknebel: how would one change that if one is matching the dictionaries directly
# 15:39 KartikPrabhu for mf2py in the tests I only check specifically what I want to test (for instance deduplication and ordering of classes)
# 15:40 sknebel Sure, can't do that. But we can structure the suite in a more feature-oriented way
# 15:40 KartikPrabhu also a direct dictionary comparision will always fail now since both php-mf2 and mf2py add a "debug" dictionary :P
# 15:41 Zegnat Oh, hmm, I do not account for that one in my test suite tester. Good call
# 15:42 Zegnat And yeah, while we can’t test specific items in the objects in a general input-output test, you can make the input as minimal as possible
[kevinmarks] joined the channel
# 15:44 [kevinmarks] both kinds of tests make sense - small synthetic example mf2 markup to test specific parsing issues, and bigger examples from the wild to be more like an integration test
# 15:45 Zegnat Agreed [kevinmarks]. But I think some of the “bigger” tests in the current suite aren’t taken from the wild either...
[eddie] joined the channel
# 15:52 [eddie] !tell skippy: There are two different things. A photo post (like Instagram style) where the photo is separate from the content. That is where a note is posted with a photos array. An article with a featured image has a featured array so the image is known to be supplementary to the post. Finally, there are embedded photos, which requires an article posted in JSON format with a content.html property that puts img tags in the correct placement in
# 15:52 Loqi Ok, I'll tell them that when I see them next
# 15:52 Loqi Ok, I'll tell them that when I see them next
[cb] joined the channel
# 16:13 Loqi skippy: [eddie] left you a message 20 minutes ago: There are two different things. A photo post (like Instagram style) where the photo is separate from the content. That is where a note is posted with a photos array. An article with a featured image has a featured array so the image is known to be supplementary to the post. Finally, there are embedded photos, which requires an article posted in JSON format with a content.html property that puts img tags in the correct placement in
# 16:13 Loqi skippy: [eddie] left you a message 20 minutes ago: oops that was three, not two :)
snarfed joined the channel
# 16:14 skippy i'm trying to wrap my head around how quill will work. if i write a long post, and supply an image, how do i get to control where that image is displayed within my post? say I want some opening text, then the image, then more text. How do I reference the image in the post when I compose in Quill?
# 16:15 loqi.me edited /Path (+94) "Zegnat added "https://thenextweb.com/insider/2018/03/22/path-ceo-considers-rebuilding-his-social-network/" to "See Also"" (
view diff )
# 16:16 [eddie] You would want to do that in the "Article" view in Quill. In that view, there is a plus icon on the left of the posting view. Wherever you press the + button, it will add the photo
# 16:16 [eddie] Then Quill sends a JSON micropub request with the <img> embedded
# 16:17 [eddie] So you would enter your text, click on the image button on the left, upload the image, then enter more text. It will send html that contains that same order
[tantek] joined the channel
# 16:18 skippy i'm still futzing with modifying nanopub to my own needs; and weighing all that effort with trying to set up Known instead.
# 16:20 [eddie] ohhh okay. nanopub, does that mean you use Jekyll or Hugo?
# 16:20 skippy though I haven't posted anything in almost a year. *sigh*
# 16:20 [eddie] ahh okay. One issue I had with the Quill Article thing is Jekyll uses Markdown, so I have to convert HTML to Markdown for my post
# 16:21 skippy Hugo accepts whatever i give it. HTML and MD. So that ought not be a huge problem.
# 16:21 [eddie] I ran into issues with html -> markdown with images embedded so I always have to manually tweak that when I do it
# 16:22 skippy but I have two different sites, with different posting requirements. Trying to determine if a one-size-fits-all micropub widget works, or if i take on the burden of maintaining separate micropub endpoints for these.
# 16:23 skippy it's all so fiddly; but i'm fussy enough to rail against the constraints of canned solutions like WP or Known.... I don't want a ton of cruft CSS or JS sent across the wire when none of my sites need any of that.
# 16:24 [eddie] Yeah, I know what you mean. I just can't do WP anymore (no offense GWG 😉)
# 16:25 skippy it takes 25 lines of custom functions.php content in a theme to strip out all the junk that WP embeds by default. Too much.
cweiske joined the channel
# 16:42 skippy can i get an easy-to-read human-readable list of all the fields that Quill HTTP POSTs to a micropub endpoint for the editor and notes?
# 16:43 skippy parsing the code is a little tedious for me, this AM.
# 16:44 Zegnat Does it send POST fields from the editor? Not just a single JSON body?
# 16:44 skippy let me rephrase: can I get a human readable display of what it posts?
# 16:45 skippy if no such list is readily available, I can capture it from a test endpoint; but seems like something that should be documented somehwere?
# 16:47 skippy thanks. but you're parsing the json, right? that's not raw, because sink escaped the content.
# 16:48 [eddie] That said, other clients might send it as a URL from a media endpoint
# 16:48 [eddie] Oh, in fact, Quill *might* send a url if you have a media endpoint
# 16:49 [eddie] I don’t think the raw, unescaped content is documented anywhere
# 16:50 Zegnat Ah. No. Sink does a little bit of stuff. E.g. when it gets a JSON body it will make sure it didn’t contain your authentication (strip it if possible) and re-render it through PHP’s json_encode with pretty printing
# 16:51 skippy ok. So the Quill "editor" and "note" links both send h-entry. And I see the format. Thanks Zegnat !
# 16:51 Loqi zegnat has 49 karma in this channel (185 overall)
# 16:53 skippy but it's not clear to me if that's matching, or if it's falling through to the json_decode(file_get_contents('php://input'), true);
# 16:55 aaronpk skippy: you might want to test with https://micropub.rocks since it will show you the fields it sends before it makes the request and has a bunch of different combinations of properties that you'll encounter
# 16:55 sknebel nanopub tries to take formencoded and turns it into the same structure as the JSON would be, or if there is no form-encoded it parses it as JSON
# 16:55 aaronpk that will make sure you don't accidentally hard-code this to only work with Quill :)
# 16:55 Zegnat form-urlencoded gets crudely converted into JSON by Sink. It should show the h property
# 16:55 Zegnat shpub is another option that can show you the full curl command before it even sends it
# 16:55 [eddie] I don’t know why I didn’t think to recommend mp.rocks that will definitely help
# 16:56 [eddie] Because it has two different tests, one for urlencoded and one for json
# 16:56 [eddie] (well more than that, but it includes both of those tests)
# 16:57 skippy oh, I see. I didnt realize that it was two different submission formats. That explains things. Thank you.
# 16:58 aaronpk yep either mf2 JSON or the simpler form-encoded format. you won't get HTML posted at you in the form-encoded version.
AngeloGladding, KartikPrabhu and [cb] joined the channel
# 17:27 swentel aaronpk, did you find a moment to checkout the gateway problem on aperture?
# 17:28 dgold oh, yes, aaronpk: My aperture was done at an early stage in your breakneck development cycle. It doesn't have a Notifications Channel
# 17:28 dgold can I trivially fix that, or should I start all over again?
# 17:29 [eddie] dgold: Should be pretty simple. We just have to add an SQL insert into the channels table for that
# 17:32 [eddie] dgold: You are the only user in your Aperture instance, correct?
# 17:33 [eddie] :thumbsup: Just wanted to make sure your user id was 1, which it will be if it’s only you 🙂
[colinwalker] joined the channel
# 17:33 [eddie] Zegnat: It’s added when a user is created, but dgold and I didn’t use the create user script
# 17:37 [eddie] You’ll just need to replace {database_name}
with the name of the database that holds your aperture data
# 17:42 dgold just need to remember my mysql password. lol
# 17:45 [eddie] lol i always look at my settings file on my server
j12t, tantek and Kyle-K joined the channel
# 19:08 sebsel is slowly chipping away at a rebuild of his site
# 19:09 sebsel Trying to get a basic working version out soon, so I can do an XXdaysOfIndieWeb :P
cweiske joined the channel
# 19:24 sebsel Laravel, since I work with that every day now and I want to practice outside of work too :) Kirby was nice, but I kept hitting walls and bugs.
KartikPrabhu joined the channel
# 19:52 cweiske is there some ready-made javascript that a not-very-technical user can copy&paste to render webmention.io data for a blog post?
# 19:52 cweiske webmention.io gives that for the streaming api, but not for the "normal" fetching
# 19:58 Zegnat Didn’t voxpelli’s webmention endpoint have some easy to copy-paste embedding code?
# 19:59 cweiske (0 of 150 accounts left at the moment – more will be released as development goes on)
KartikPrabhu and hurdygurd joined the channel
# 20:43 aaronpk wow i really broke something with aperture's paging mechanism in that last update
tantek and snarfed joined the channel
# 21:17 cweiske turns out that css grid is awesome. no floats anymore
[jeremycherfas], snarfed and chrisaldrich joined the channel
# 21:31 cweiske I always thought "u-" to define the data type as URL
# 21:31 tantek cweiske: that's a common misconception of the mf2 prefixes. the prefixes are for parsing differences
# 21:32 cweiske but the rule that u-comment requires to look for an u-url inside looks strange
# 21:32 tantek and yes those parsing differences are adapted from authoring use which often happens differently for different properties on different elements
# 21:32 tantek it's part of how property nested microformats objects are handled
# 21:33 tantek e.g. on something with class="u-comment h-cite"
# 21:34 Loqi zegnat has 50 karma in this channel (186 overall)
# 21:35 Loqi [Zegnat] #163 Run mf2/tests test suite with PHPUnit during default testing
snarfed, [kevinmarks] and dentonjacobs joined the channel
[cb] and chrisaldrich joined the channel
kaushalmodi, snarfed and KartikPrabhu joined the channel
[tantek] and snarfed joined the channel
eli_oat joined the channel