vika_nezrimayaMy own website could be an IndieAuth client to, e.g. authorize users for viewing private posts and to provide embedded editing/deletion functionality when the website owner (i.e. me) is logged in via IndieAuth
jackybut there's a chance that if I set it to be published in 30 seconds that the _request_ takes 40 seconds (somehow, my requests tend to complete in less than 2) and it ends up being published before I can get a chance to change it
Seirdyi also don't think voiceover is the only one. if it is then this could be considered a proprietary quirk and not an issue that everyone should cater to...
petermolnar, hoenir, hans63us[d], yequari[d], edburns[d], Silicon[d], shaunix[d], wackycity[d], indieweb-irc-bri, laker[d], marksuth[d] and jamietanna joined the channel
jamietannaIf anyone fancies doing some IndieAuth testing, https://gitlab.com/jamietanna/micropub-go now has an `auth reauth` command which should automagically go through the auth flow using an OAuth2 client, but it'd be good to know if it does work across various servers
jackylike at first like at first, I thought about wrapping the whole phrase "Jane F. Doe's" in a h-card but wrapping the name in a <span class="p-name"> (or whatever works for you tag-wise) and keeping it all in the link so semantically it's only the name that's marked up but I _think_ a screenreader would read those as the same unit _within_ the link
sknebel(authoring-wise, e.g. markdown will turn [[Webmention]]s into <a href="Webmention">Webmentions</a>, so one could extend a tag syntax to automatically include 's along those lines)
ZegnatThe other question is if that specific instance of a person’s name must be linked. Is that the click target for the whole of the content? Is there something else that is a better click target? So rather than rewriting the text as jeremycherfas says, it might be about restructuring your linking instead.
capjamesg[d]In hindsight, maintaining my infrastructure would be easier if everything was reverse proxied to a single domain because I wouldn't have to maintain eight or so SSL certs and config files.
Caesar[m]capjamesg[d]: why not a wildcard cert? I use subdomains (and also more than one root domain but that's another story) and never had an issue with managing certs with LetsEncrypt. I do use reverse proxying to local sockets/ports (with Nginx) but from the subdomains.
jamietannacapjamesg[d] alternatively 1000% recommend looking at caddyserver.com as Caddy is great, super easy for SSL setup and you don't need to think about it after the initial setup :)
Caesar[m]jamietanna: thanks for reminding me of Caddy, I must have a proper play with it sometime. It looks great – not that I've ever had any complaints about nginx.
Loqi[tomwassenberg] certbot-ocsp-fetcher: A tool that primes the OCSP cache of nginx for certificates managed by Certbot, in order to make OCSP stapling work reliably.
Seirdyi use those with a statically-linked nginx compiled with boringssl. can handle over 1.4k req/sec. caddy could do like 950/sec. My biggest DoS was 1.2k/sec. so i should be able to handle a low- to medium-level DoS attack without much rate-limiting.
ZegnatThough that has been because I was trying out docker for things. Traefik can keep a look at all your containers, and can take settings from docker labels. So I can add a new application image to docker, set a hostname and port in the docker labels, and Traefik automagically hooks me up with access and https
sknebelyeah, that was more a "ok, they supply a readymade traefik config and the server isn't doing anything else anyways... guess I'm trying traefik for this" for a specific setup
@MuseResortOur #WritingPrompts website now supports #WebMentions which means you can notify us of a story based on one of our prompts. Please do share your writing - we would love to read it. (twitter.com/_/status/1513244169992642564)
[arush]Hi all. Further back in threads, someone asked if screen readers require action for a span. The short answer is that it depends on what's in the span. If the span has something like onclick attached, then keyboard listening would also have to be attached because screen readers and switch mice and dictation software rely on operability with the keyboard not mouse so that needs to be accommodated. If you're doing something like using a
[arush]span to override the lang attribute for a specific piece of text that's in another language other than the page's primary language no ARIA is required.
Seirdy[arush]: if you're able to see scrollback, I linked a resource from Axess Lab about this. the ARIA "text" role works for telling ATs that enclosed spans are not semantically significant.
[arush]I saw the accesslab link and will bookmark that, but spans already don't have semantic significance so I'm wondering why they're suggesting an Aria fix when rule 1 of Aria is don't use Aria until you have to. They probably have a specific use case in mind though.
Caesar[m]The `[hyperlinked name]'s` issue seems like a bug in the screen reader, the semantics are clear both from the markup point of view and the English language point of view. But (speaking as someone who doesn't have to use one) every time I try testing a site with VoiceOver I realise how terrible VoiceOver is…
SeirdyCaesar[m]: Apple argued it's a feature not a bug. also happens in Orca even with the role=text. i think it's best to include the apostrophe-s inside the hyperlink and to put the p-name in a span inside the hyperlink, separate from the apostrophe-s. that causes orca to read it correctly and it honestly seems better as a sighted user too.