#dev 2025-04-04

2025-04-04 UTC
#
carrvo I'm glad to have sparked some cleanup!
cow_2001, sebbu2, webber3, barnaby, a12r, geoffo, sp1ff and [Jo] joined the channel
#
doesnm gRegor: after git.0ut0f.space was alive my results more perfectly https://indiewebify.me/validate-rel-me/?url=https%3A%2F%2Fdoesnm.cc
#
doesnm not working only 4pda
epoch and DusteD joined the channel
#
[tantek] capjamesg, when I add to Home Screen your /contact page (which as said redirects to /email), and then tap to open it, it launched in its own process and redirects to your home page
#
[tantek] Both of those seem unexpected behavior for a /contact page
#
[tantek] Was that intentional?
#
capjamesg Hm. Weird.
#
Loqi capjamesg: [tantek] left you a message 10 hours, 34 minutes ago: can you move create a PR to move your implementation report https://github.com/w3c/webmention/blob/main/implementation-reports/webmention.jamesg.blog.md to the proper folder: https://github.com/w3c/webmention/tree/main/web/implementation-reports ?
#
capjamesg I will take a look!
barnaby, claudinec, nemonical, nemonical1, ttybitnik, duanin2, dsf and fdgg joined the channel
#
zachary.kai Think I decreased the loading time of my site significantly! Thanks to what capjamesg[d] and artlung[d] have mentioned in previous HWC (the picture element: offering .webp and png/jpg for images and the @import feature for the css stylesheet.) Thanks folks!
#
artlung awesome! nice job zachary_kai[d]++
#
Loqi zachary_kai[d] has 3 karma in this channel over the last year (4 in all channels)
#
carrvo https://github.com/w3c/webmention/pull/119
#
Loqi [preview] [carrvo] #119 SVNmentions v1.0 implementation report
#
carrvo [edit] https://github.com/w3c/webmention/pull/119
#
[artlung] Yesterday I encountered a site which had its webmention `link` rel tag added using JavaScript. Are there are any folks or implementations which send webmentions whose processes would catch this and _still_ send?
#
aaronpk i doubt it
#
capjamesg Mine wouldn't.
#
capjamesg I do a GET request without JS rendering.
duanin3 and duanin2 joined the channel
#
[tantek] artlung that wouldn’t pass the test suite
#
[artlung] It reminds me of an anti-spam measure. In this case it was because most of the (very nice and I won't call out by name) site was js dependent.
#
carrvo Is there an efficient way to use php-mf2 to verify that a source has truely webmentioned a target?Ideally this would focus in on the item that has the Webmention so that I can inspect the surrounding properties.
#
carrvo Currently I am doing the simplest plaintext search...
shoesNsocks joined the channel
#
[tantek] carrvo “truly” depends on your specific use cases for receiving webmentions. I’d suggest documenting those use-cases first, then that will drive what will or won’t work
gRegor joined the channel
#
carrvo Whether there is any mention at all? That is where I am starting: accept all and treat them the same. Then expand to differentiate.
#
carrvo But having h-* inside h-* to any depth seems like trouble...
#
gRegor I personally wouldn't. I think the only way it might be slightly useful is if you wanted to limit which part of the page the link is in, like the parsed `content` HTML. You'd still be matching text, though, and it would require a site have mf2 in order to successfully mention your link.
#
gRegor My webmention plugin parses the DOM to extract links. It's also PHP if you'd like a reference: https://github.com/gRegorLove/ProcessWire-Webmention/blob/main/Webmention/Webmention.module#L631
sebbu2 joined the channel
#
sebbu i finished making my own imgur backup (with albums data & all), which script should i selfhost to replace it : immich, nextcloud, pixelfed, something else ?
#
doesnm e949 xD
#
doesnm just joke, don't try to find it
#
sebbu "hydrogen when used as packaging gas"
#
sebbu i found an e621 but it's probably not that (don't try to find it, nsfw)
#
sebbu https://github.com/e621ng/e621ng but it would be possible :p
#
carrvo gRegor, does that mean I can assume Webmentions links are exclusively within a predefined set of tags (link, a, img, et cetera)? That would certainly simplify things.
nemonical joined the channel
#
sebbu form, input, button, area, span, div
#
[tantek] There’s a lot of href and src attributes and could be more in the future
#
[tantek] like audio and video were added. Then the source element. Then the picture element etc. HTML is a living standard
#
[tantek] gRegor++ for the DOM approach because that’s a lot more future proof. Let the browser parse for you instead of scraping the page
#
Loqi gRegor has 32 karma in this channel over the last year (104 in all channels)
#
doesnm sebbu: https://git.projectsegfau.lt/Cyclone-Team/e949
#
doesnm look at dev branch
#
gRegor carrvo, It's a "SHOULD" in the spec to use media-type rules to verify, so it's mostly up to you. Personally I'm fine with assuming text/html for now and checking `a` elements. that is older code and I'm slowly updating it so might expand it some
#
gRegor https://webmention.net/draft/#webmention-verification
#
gRegor Another advantage of the DOM approach is that I can send myself webmentions using relative links. That code snippet resolves relative URLs and normalizes the hostname too, in case any weird people out there might used mixed-case when linking their domain ;)
#
gRegor I recall someone working on verification for webmentions received from a JSON document, though don't know if that's in prod anywhere or was an experiment.
#
carrvo Good point. I also want to Webmention myself as a form of tag aggregation.
#
carrvo gRegor++ you have given me good thoughts for a next version!
#
Loqi gRegor has 33 karma in this channel over the last year (105 in all channels)
[marksuth] joined the channel
#
[tantek] 13:37 time to hack something
#
gRegor the planet!
#
Lars-Christian @sebbo immich looks sweet.
GuestZero and thegreekgeek joined the channel