#dev 2026-02-02
2026-02-02 UTC
gRegorLove_, ax562, jjuran_, bterry1 and grufwub joined the channel
# 
khurtwilliams php-mf2 parser → succeeds on my actual page, returns all 3 top-level objects in order
# khurtwilliams gRegor: I think I've isolated the issue. XRay is returning "type": "unknown" because my post (https://islandinthenet.com/indieweb-2030/) has a standalone h-card in the header (line 113-128 of the HTML) that appears before my h-entry article in DOM order.
# khurtwilliams My actual page with h-card first → XRay fails ("type": "unknown")
# khurtwilliams According to microformats2-json spec: "items is an array of microformats2 objects, ordered according to their order in the source document."
# khurtwilliams Created minimal h-entry test page → XRay succeeds ("type": "entry")
# 
IWDiscord <khurtwilliams>
# 
Loqi khurtwilliams: gRegor left you a message 7 hours, 19 minutes ago: I think there might be a caching issue on https://islandinthenet.com/indieweb-2030/. Also I think the comments on your post should be in `u-comment h-cite` so they show up as children of the entry. You can try pasting in variations of the HTML on https://xray.p3k.io/ to see how it parses. IndieNews is using XRay under the hood
# IWDiscord <khurtwilliams>
# IWDiscord <khurtwilliams>
# khurtwilliams So parsers should return ALL top-level microformats, not stop at the first one.
# IWDiscord <khurtwilliams>
# khurtwilliams Is XRay's behavior (stopping at first h-* object) intentional, or is this a bug? The standalone h-card is redundant anyway since I have an h-card inside the h-entry as p-author, so I can remove it. But I want to understand if XRay is following the spec correctly.
dale and dale-- joined the channel
# dale-- I'm trying to indieauth on to the https://xn--sr8hvo.ws/ web ring centre, but it is failing without telling me why. Is schmarty here by any chance?
# dale-- Anyone associated with that site?
# dale-- khleedril.org is my domain.
# dale-- I'll have to call back in 8 hours then! Thanks.
# 
gRegor You can try entering that second link in https://indielogin.com to sign in. Your site shows an "OAuth(DenySilently)" message then some JSON when I enter that
# dale-- I'm not currently recognizing that client-id. Give me ten minutes, I'll add it in my code.
# gRegor From the webring, I entered https://khleedril.org/dale-mellor and got "The authorization server did not return a valid response
# gRegor That message indicates (at least) no "me" in the response JSON from your server, from the client library: https://github.com/indieweb/indieauth-client-php/blob/main/src/IndieAuth/Client.php#L135 Might check your logs for what response it's sending.
# dale-- Yep, I'm on it. Thanks for your help.
KE0VVT, rolandas, simpleauthority, jjuran, blacklight and [simonbc] joined the channel; dale-- left the channel
# [simonbc] I'm starting implementation of IndieAuth for Jottit today. Jottit authenticates using magic links. Has anyone else done this combo? Any gotchas?
rolandas joined the channel
# 
[KevinMarks] Indielogin does email a passcode, which is similar. The usual magic link issue of having 2 tabs open on your site, sometimes with one inside the users email app on mobile applies.
rolandas joined the channel
# [KevinMarks] Interesting PhD job in implementing decentralized social https://hci.social/@floe/116001038629671413
[schmarty] joined the channel
# 
[schmarty] gRegor++ thanks for the webring auth assist! i don't know how to karma dale, who has a -- at the end of their name. 😂
ax562 and rolandas joined the channel
# [KevinMarks] Oblique dev method: https://fosstodon.org/@kitten_tech/116001479944727697
barnaby joined the channel
# [simonbc] KevinMarks++ Thanks for the heads up about the two-tab issue! When you say IndieLogin emails a passcode - is that a code users type into the browser, or a link they click? Trying to figure out the best UX for magic links + OAuth.
[jeremycherfas] and blacklight joined the channel
# 
WinnerWind ive been thinking of how one could use a LAN as a way to create physical instances of docker.... let me explain what i mean
# WinnerWind usually what people do is that they install docker and run containers within a system, which are kind of like virtual machines
# WinnerWind however, what if we had one server doing a reverse proxy, and other physical servers each tasked with running only one service, and that reverse proxy server just goes "yeah i'll proxy this IP address to you which I know is in my LAN"
# WinnerWind it's a stupid idea but it could be useful for things like....... jellyfin and stuff where you probably dont want the power hungry server to be on all the time
# [KevinMarks] Aaronpk may have more up to date advice on magic links and OAuth, I think there were some recent specs to clarify some of the paths
[Schepp], thegreekgeek, blacklight, KE0VVT, rolandas, rrix, [Joschi_Kuphal], ax562, oodani, sebbu, mahboubine, [snarfed], btrem and barnaby joined the channel
# btrem Except for a brief period when I first learned javascript (I don't want to date myself, but it was a *long* time ago), I've always avoided the <noscript> element, opting instead for progressive enhancement. But the popover api makes me wonder if a <noscript>-ish use of the api would be acceptable?
# btrem I have a page with a button that copies a public key if javascript is enabled. In the markup, the button is hidden and there's any empty popover paragraph. The javascript unhides the button, and, when clicked, copies the key text. It also changes the popover message to "Text copied" if it worked, or an error if it didn't.
# btrem Since I'm modifying the popover text anyways, I could change the html markup, leaving the button visible, and in the popover element writing something like "The copy button is not working. Maybe you have disabled javascript in your browser?" Then, if javascript is disabled, clicking the button will inform the user that nothing happened.
# 
[artlung] Brennan Day's recent article is great on a rationale and method for noscript: https://brennan.day/respecting-the-no-js-choice-making-your-site-work-for-everyone/
# gRegor btrem, I've built a "copy link to these results" UI which with JS enabled, just shows a button that copies to the clipboard. With JS off, it shows a text form element with the link in it, label "Copy link to search results". I'd prefer this type of setup than any warning about something not working and needing JS.
# gRegor With https://developer.mozilla.org/en-US/docs/Web/CSS/Reference/Properties/user-select, you can also make selecting that text field select the full contents. I think [tantek] just did that recently on on the citation boxes, like https://tantek.com/2026/028/t1/csswg-f2f-apple-cupertino
# btrem It's mostly a question of what to do with the button element. Option (a) hide it by default, enable it in the js. Option (b) attach a popover message explaining why it doesn't copy. The idea of a <label> to change the focus to the text to be copied is clever, but might not be workable for a long public key.
# gRegor I do something like that with my Cryptpad key here: https://gregorlove.com/follow/#cryptpad Have not considered building a copy UX before, since Cryptpad is such an edge case, haha
[Al_Abut] joined the channel
# [Al_Abut] [KevinMarks] I definitely concur that the challenge of making decentralized social media usable by laypeople is a PhD level of hard.
# [snarfed] I dunno. that's often true - fediverse, Nostr, SSB, etc - but not always. Bluesky and http://micro.blog are imo two counterexamples
# [Al_Abut] [snarfed] fair point!
# [Al_Abut] I often refer to http://micro.blog and omg.lol as the “solved” versions of a beginner-friendly on-ramp to the indieweb. All of the building blocks are there as features to be used, not specs to be read.