KartikPrabhu, [prtksxna], [xavierroy], doubleloop, mblaney, [tonz], [Michael_Beckwit, cweiske and sonniesedge joined the channel; prtksxna left the channel
#jamietanna[m]Finally been making good progress on my Micropub endpoint! Got it committing straight to a GitLab repo via the api - need to do some more validation on request, and then hook in tokens, but then it'll be up woo!
KartikPrabhu, [fluffy], [xavierroy], Loqi and [jgmac1106] joined the channel
#GWGRereading the webmention spec and pages on the wiki
#ZegnatComing to any conclusions, GWG? Do you have a goal?
#Loqi[Zegnat] php-linkextractor: Class for finding all resources an HTML document links to.
#GWGBut the feedback was having it as a separate plugin was confusing
#GWGI plan on using ideas from your link extractor
#ZegnatI guess it is a usecase thing. I do not currently extract content from pages that webmention me at all. Not in a separate process either.
#GWGOne of the changes I want is to use DOMDocument instead of regex
#ZegnatMost I do (though I have temporarily stopped that at the moment) was feed pages that validated into a full web archiver that I ran. So I had a copy of the full page for future reading.
#ZegnatBut I guess people who want Webmentions in WordPress will almost always want to combine that with extracting data from those pages. So then merging the processes makes sense
#GWGThe code uses regex to verify links, then later parses using DOMDocument
#GWGI want to move the creation of the DOMDocument up further and pass it between functions
#ZegnatThis is also not about the split between sending and receiving, but between receiving and parsing. I am not sure I have seen many platforms that split receiving and parsing
#[jgmac1106]just had another random idea, could social readers be away to think about webmention moderation?
#[jgmac1106]I already get notifications of webmentions (in my CMS and from webmention.io)...why not a button for "Display webmention"
#ZegnatSure, you could subscribe to a feed of your own webmentions in your reader.
#ZegnatStill just notifications though. I am not sure it makes sense to add moderation tools inside a generic reader. But I am the wrong judge for that, I use an oldschool reader ;)
#[jgmac1106]yeah makes more sense in my CMS...just thinking about a "social reader" as a "classroom" for a bit and wanting to give folks control of when or when not to display
#[jgmac1106]when using WP it isn't a problem as askimet does the moderation
#[jgmac1106]for fall people will have to manually send and receive using telegraph and webmention.io...I bet most won't do that step
[eddie] joined the channel
#[eddie]I've thought about webmention moderation from within a reader
#[eddie]I came to the conclusion that the feed would need to be private and then I could display links in the "post content" at the bottom for moderation
#[eddie]that way the CMS is sending the actual links, and you are just clicking on links within your reader. It would work with both social readers and old school readers
#[jgmac1106]yeah eddie I was thinking about that I was copying and pasting a few webmentions
#[eddie]I don't currently moderate my webmentions, but as I've thought about it, I think that's the first thing I'll do towards it
#[eddie]I think that's my primary plan. But the other thought I had was in terms of making all webmentions available through some Micropub Query
#[eddie]so then a Micropub client could query them and adjust their visibility and status attributes
#[jgmac1106]that is what I was wondering could be the "display webmention" action, its held in moderation until published with micropub to your site
#[jgmac1106]also want to add a button that would remove the webmention endpoint from any post so I can toggle them on and off
#[eddie]ohhh interesting. So while you were creating (or editing) a post, you could essentially "turn comments on or off"
#[jgmac1106]yeah just a small script to remove the endpoint, that would be an important feature if I used webmentions in school settings
#[jgmac1106]prevent dogpiling on a specific post by bad actors
#ZegnatWhy would you stop advertising the endpoint rather than a/ blocking on receipt or b/ not displaying?
#[jgmac1106]I dunno just seemed like an easier...deleting one line...
#ZegnatI think some senders (IIRC, this includes bridgy) will cache webmention endpoints on a per domain basis. So you not advertising it on one post will not affect them, they will continue to send for that post as well as every other post
#ZegnatWow, sebsel, since when do you syndicate long form blog posts to Twitter threads?! I thought you had a bug when your latest post displayed 13 syndication links, but apparently not
[schmarty] joined the channel
#[schmarty]Zegnat: caching is only part of the spec insofar as respecting response cache directives. Also, removing the endpoint header prevents new senders from finding it.
#[schmarty]It would also reduce load on a webmention receiver to... not receive mentions 😁
#Zegnat[schmarty]: oh for sure. And caching-per-domain is probably actually a spec violation. I am just saying that it is how some senders work today.
#ZegnatSo if you see the majority of your mentions come in from bridgy, just removing the header is not going to do much
#ZegnatI’d be in favour of both removing the header in addition to handling it on the backend.
#Loqimultipleoverlappingapproaches has 1 karma over the last year
#ZegnatAlso: you probably already have some sort of logic to ignore webmentions that aren’t for your website in the backend. Extend that to also be able to silently drop URLs for specified articles on your domain and you are there
#[schmarty]I maintain that it's better not to receive an unwanted webmention than to have to handle bad ones
#ZegnatI agree on all points, [schmarty]. Just want to warn people that it may not be the actual way things are set up today :)
#Zegnat[Also insert note on how lucky we are we do not have spam problems yet here]
#cweiskenote to self: build a tool to send spam webmentions to all irc-people
#LoqiChaos Monster is an opt-in IndieWeb project by Eddie Hinkle that aims to help IndieWeb resiliency similarly to how Chaos Monkey helps Netflix https://indieweb.org/Chaos_Monster
#[eddie]Chaos Monster << idea: for those that opt in? Send spam webmentions
#[eddie]Chaos Monster << For those that don't opt-in but are obviously part of the IndieWeb, (like irc-people) consider sending a single spam webmention to prompt them to consider their webmention workflow, and then don't send any more unless they opt-in
#Loqiok, I added "For those that don't opt-in but are obviously part of the IndieWeb, (like irc-people) consider sending a single spam webmention to prompt them to consider their webmention workflow, and then don't send any more unless they opt-in" to the "See Also" section of /Chaos_Monsterhttps://indieweb.org/wiki/index.php?diff=64035&oldid=64034
[Jacob_Michelsen, [Rose], [davidmead], eli_oat, [tantek] and KartikPrabhu joined the channel
#[tantek]For those interested in IndieWeb issue tracking, starting by owning your own issues that you write!
#Loqiissue is a special kind of article post that is a reply to typically some source code, though potentially anything at a source control repository https://indieweb.org/issue
#[tantek]And add yourself to the IndieWeb Examples there
#[jgarber]“Space-separated” in URLs could be the plus sign or `%20`. What’s being used in the wild in this specific case of passing `scope` values to an IndieAuth server? I’m seeing the plus sign used more than `%20`, but should the server attempt to account for both? Or more possibilities?
[kimberlyhirsh] and [jgmac1106] joined the channel
#[jgmac1106][Jacob_Michelsen][tantek] I wonder if people are grabbing different themes and then editing them and losing some mf2. I just checked a few sites hosted on micro.blog and you get correct, some, and none with mf2
#[jgmac1106]or the main h-feed doesn't have h-entries unless you click on the link, seen that a few times as well. think folks are just grabbing Hugo themes
#[tantek]hmm looks like when I added alt text to my images I broke syndication to micro.blog — none of my photos are making it through: https://micro.blog/t
#[tantek]!tell [manton] when you have a moment, it seems we need to better document (or link to where it is documented) how to syndicate /photo posts to micro.blog. E.g. by adding alt text to all my photos it seems like I've completely broken syndicating my photos to micro.blog https://micro.blog/t 😞 — not your fault, I just need to know how to fix it, and I'll document that on /micro.blog too
jackjamieson joined the channel
#[tantek]Thanks to [gerwitz] for the heads-up BTW.
balupton[m]1, kitt, eli_oat, jjuran and jackjamieson joined the channel
#[jgmac1106][tantek] [Jacob_Michelsen] FYI you are correct I tried two default themes in micro.blog and the mf2 is messed up
#[jgmac1106]for example the Arabica theme a post has the correct markup but the h-feed and the h-entries on the homepage are missing
#[jgmac1106]already pinged to find out where the repo is for the themes to see if I can fix
#[tantek][jgmac1106] so that should be something we can fix in microblog and then hopefully add tests to make sure they don’t break again
#[jgmac1106]yes, I also suggested in the Slack we add a paragraph about mf2 in the theme customization tutorial, many people just grab off the shelf Hugo themes
#jgmac1106themes and plugins all the flexibility and all the maintenance….
#[tantek]Can you add a link to the IndieWebify.me h-entry validator too?
#[tantek]The key is making sure the code repo has continuous integration with tests that pass / fail so breaking changes can be backed out quickly and fixed
#[tantek]That significantly reduces the maintenance pain
#Zegnat!tell jgarber interesting issue with the plus sign. I am not sure those are considered spaces at all by the WHATWG URL spec that IndieAuth refers to. %20 may be mandatory. Open an issue for discussion?
#ZegnatOh, interesting, does any Jekyll/Hugo (whatever microblog was based on) theme work, jgmac1106? I can totally see why people end up picking ones without mf2 then
#[jgmac1106]yes technically, but it seems most versions I am checking use the default themes you can choose, going through and testing those now
#[jgmac1106]will add what is missing to the wiki and then go back and fix them when I ever I figure out where the repo is
#ZegnatI don’t think [manton] made any of that publicly available? Or he may even be using the actual repos of thsoe default themes rather than his own forks? You’d have to ask him
#jgmac1106most of the default themes are just open sourced Hugo themes
#Zegnat[jgmac1106]: may be better to document missing mf2 for a theme in an issue on the theme’s repo where people in the m.b ecosystem can find it, rather than on the wiki
#jgmac1106yeah…I will do that once I figure out correct repo, really notes for me when I go back and try to fix
[jgarber] joined the channel
#[jgarber]Zegnat: Yeah, there’s the double-edged sword of “plus signs aren’t whitespace” and “well crap plus signs are used as whitespace characters in URLs across the Web”… 😭
#Loqi[jgarber]: Zegnat left you a message 15 minutes ago: interesting issue with the plus sign. I am not sure those are considered spaces at all by the WHATWG URL spec that IndieAuth refers to. %20 may be mandatory. Open an issue for discussion?
#Zegnat[jgmac1106]: check sknebel’s link. I think the repo names will match whatever names you are selecting within m.b
#[jgarber]FWIW, looks like several of [aaronpk]’s apps (Aperture, Monocle) use plus signs to pass around scope values (which would be in accordance to the example in the spec, at least…).
#[jgarber]Not sure how anyone else is passing scope values around…
#[snarfed]comma and plus are both common scope separators
#[jgarber]snarfed: Oh, goodness. Comma??? Where have you seen that in use?
#Zegnatspace U+0020 is the official separator in OAuth, IIRC. Comma should not be used here.
#[snarfed]silos. can find examples if you need them.
#sknebelZegnat: WHATWG URL spec doesn't say anything about query part of an URL afaik outside of "it's a string, here's a list of characters it can have, everything else must bei percent-encoded"
#[jgarber]…maybe I’ll split on every non-alpha character. 🙃
#[jgarber]Of course that’s only as good as what folks are implementing in the wild…
#Zegnatsknebel: darn, you are right and I am being dumb. The spec describes how to break out the query part of the URL, but not how this is deserialised into something other than a single string
#[jgarber]Zegnat: So plus signs only then for IndieAuth things…?
#ZegnatI do wonder what URL spec we are supposed to use when the IndieAuth spec writes “adding the following parameters to the query component”. If that is not specified behaviour of the WHATWG URL spec
#Zegnat[jgmac1106]: either + or %20. We need to figure out which spec to refer to for encoding the query component -.-'
#[jgarber]Trouble is the spec says one thing in prose (“space-separated list”) then demonstrates a different thing (plus signs) in the example…
#[jgmac1106]fragmention discussion goes back and forth on this
#[jgarber]☝ That’s the recent conversation I was trying to remember!
#ZegnatNo. The demonstration is *after* serialising into a URL. If the query component serialisation defines plus signs for spaces, the demonstrations in the spec are correct
#ZegnatThe question is: where are we getting the serialisation algorithm from
#ZegnatAnd actually, it wouldn’t surprise me if that is the spec it is taking from. Because we often follow what browsers would do when submitting regular forms.
#sknebelseems so. but the parsing half is probably more important
#Zegnat[jgarber]: I’d do what the WHATWG spec does there for URL parsing. Take the entire query string, replace all + with literal spaces before percent-decoding, then use only literal spaces as scope separators.
#ZegnatBut may still be worth a clarification from aaronpk. I need to check on my dinner now, but am happy to file an issue after :)
[fluffy] joined the channel
#jgmac1106in the states they have launched commercial free included with youtube music….they are upping the annoyance factor to drive conversion goals me thinks
NinjaTrappeur, [davidmead] and [Jacob_Michelsen joined the channel
#aaronpkat one point, the oauth 2.0 spec used commas as a scope separator, but that was back in like 2012. unfortunately a few high profile APIs implemented commas before the separate was changed to spaces, so you still see some of that in the wild
#aaronpkbut technically OAuth 2 is very clear on space separated scopes
#aaronpkbut then you have to URL encode them to use them in a query string, and URL encoding allows both + and %20 to represent spaces, so you see both
#[jgarber]When requesting a token in the IndieAuth flow, in the initial POST request, the spec mentions `grant_type=authorization_code` being passed along with the `code`, `client_id`, `redirect_uri`, and `me` params.
#aaronpkit's the only grant type we use in indieauth right now, so technically not needed, but that's required by OAuth 2.0 because there are other grant types
#[jgarber]Gotcha. Thought it might be an OAuth thing. So were it missing in the POST, that’d be a 400 error?
#j605thanks for making indiewebify. that made the process much easier
#[KevinMarks]I made the indie-Tufte one a while ago, and adapted one recently, though I changed it enough visually that it is a fork now. Worth adding to the Hugo themes collection?
#j605I don't know but it will great to have references to mf2 implementations
[jgmac1106] joined the channel
#[jgmac1106]think we put the author h-card in the post footer and many of the themes will be fixed, I will look at the templating tonight if I get a chance
ORAORAORAORAORA[, [Jeannie] and [Jacob_Michelsen joined the channel
#[snarfed]and actually for doing this in the wp wm plugin, it also shouldn't _accept_ wms on those pages, so it doesn't matter if bridgy caches the endpoint, its attempts to send would still fail