#dev 2023-12-28
2023-12-28 UTC
[KevinMarks], ovid and trwnh joined the channel
# 
tenshi.dev I have a question for the webmention protocol. Often times the person who would like to know about the mention is the target (aka. original page owner). However, expecting every source to actively push the webmention is tricky (eg. the source site like mine, might be a static site, the owner might not be aware, etc).
# tenshi.dev What is your opinion for a webmention sending mechanism based on crawling sites for links? I imagine a separate program that would asynchronously scrape pages, search for the links, checks for the webmention capability of the target, and send the mention. Is this something that is encouraged (it seems to be allowed by design, is there an existing project that does just this?), or consent from both source and target
# IWDiscord <tenshi.dev#0>
# 
Loqi Telegraph is an open source API and service for sending webmentions, part of the p3k suite https://indieweb.org/Telegraph
# tenshi.dev I see, yeah I'm aware of the manual mechanism of sending webmention. However it requires additional effort to be made from the source side. Just wondering the stance and implications of having an automatic discovery mechanism (crawling) to support automatic webmention delivery, where the source might not even be aware of such mechanism.
[tantek] and ovid joined the channel
# tenshi.dev gRegor: nope, letting it run loose and crawl any site. I could provide seed sites, eg. those who enroll to receive webmention, then follow scrape every link those seed sites have, so on and forth. Could also be limited to certain platforms and such (eg. those in a particular webring).
# tenshi.dev oh the bridgy use case is interesting. For me personally I'm going to propose this mechanism in my community, I think we could use some interactivity across our blogs and static sites. most of us use handcrafted html static sites, most are not technical sort of person, i personally use Hugo for my site (https://rai.so)
# tenshi.dev (i think I could reuse webmention for the receiving part)
# [snarfed] http://tenshi.dev it's a good idea! at least I thought so. I did it a bit ago for Reddit and Hacker News, large scale, across historical posts and comments. I got much more negative feedback than positive so I eventually stopped. https://github.com/snarfed/hackermention
# [snarfed] if you wanted to do it web-wide, you could use something like https://commoncrawl.org/
# [snarfed] lots of overlap there with https://indiemap.org/ . it's old, but you could use that to enrich and seed the set of wm receivers. still a ton of network and compute though
# tenshi.dev [snarfed]: that's one interesting project and yes it seems really similar to what I'm trying to achieve, perhaps just a different scope! If you'd like to share, could you elaborate on the negative feedback that you received? I'd imagine there would be issue with spam and such, but it might be something else I haven't foreseen
# [snarfed] also re aaronpk's telegraph, https://telegraph.p3k.io/superfeedr is a related idea, looks for links to a specific site from anywhere and sends wms for them
# tenshi.dev [snarfed]: I see, yeah this is something I have thought about. Although I'd say it's more on the receiving side to make such policy rather than cutting out the source (same as muting mention on discord - it doesn't prevent anyone from mentioning you in the first place). But I could see why it could be unpleasant for a popular blog with decent amount of traffic
[chrisaldrich], geoffo, CRISPR, ovid, AramZS and [jacky] joined the channel
# capjamesg tenshi_dev[d] I made https://github.com/capjamesg/hn-webmention to poll the HN API and send me a Webmention to say my site was posted to HN.
# capjamesg This script only sends webmentions for my domain. This is in contrast to [snarfed]'s script which aimed to send webmentions to everyone.
# tenshi.dev capjamesg[d]: thanks for sharing, much appreciated 😄
# tenshi.dev I think in a sense my use case would be somewhat in the middle - not exactly everyone, but maybe spanning a few webrings. I'll try to make a PoC for my site
# capjamesg Yay!
# capjamesg I love your website by the way tenshi_dev[d] !
# tenshi.dev capjamesg[d]: thanks 😄 I love your writing on encouraging personal pages on the web as someone who's recently influenced by this thinking (otherwise I won't be here duh).
# tenshi.dev hopefully I can write consistently, would certainly be a challenge 😄
# IWDiscord <tenshi.dev#0>
# capjamesg Why thank you!
# capjamesg Consistently writing is a challenge.
# capjamesg I wrote no posts between the end of October and the end of November, then I wrote lots of posts in December.
# capjamesg I have tried to write every day in the past but I find I'm at my best when I instead focus on writing as soon as I have an idea.
# tenshi.dev capjamesg[d]: I have a feeling that I'm going down on the same route - I'm absolutely horrible at trying to do something consistently for the sake of being consistent hahah
# capjamesg When I try to force myself to write, I sometimes have an idea. But most of the time I end up feeling bad about not being able to write.
kleb, gRegor and geoffo joined the channel
# 
[KevinMarks] tenshi - something I do with mention.tech is both send the webmentions and store them there for the sites so it will verify the mention and store it without the site having an endpoint, as well as being an endpoint if you want it to be. That kind of approach might suit your static site group, as they'll need an external webmention provider anyway.
# tenshi.dev [KevinMarks] agreed, a platform that handles both might provide better UX for the users. And yeah definitely they need to have some external provider
# [KevinMarks] mention.tech has crappy UI, but you could fork it to do what you need
# tenshi.dev thanks! much appreciated 😄
[tw2113], AramZS, [schmarty], gRegor, [pfefferle] and [0x3b0b] joined the channel
# 
gRegor [manton], fyi I think the h-entry classes on your homepage got dropped, the properties from each post are grouping together under the h-feed: https://pin13.net/mf2/?url=https%3A%2F%2Fwww.manton.org%2F
[manton] joined the channel
# capjamesg is thinking about what to build.
geoffo joined the channel
# 
[tantek] [snarfed] re: examples 1-5, I think #4 with the full table is helping best communicate the full extent of things possible in a good summary: https://20231226t135650-dot-bridgy-federated.uc.r.appspot.com/ though I would add at least cell/heading borders to make the cells more clear, otherwise it's a bit hard to follow where text begins/ends that applies to anything in particular. Also a white background on the table to draw it out from the
# [tantek] https://chat.indieweb.org/2023-12-26#t1703630842812500 for reference for #4
# Loqi [preview] [[snarfed]] 4) table with all protocols (unstyled): https://20231226t135650-dot-bridgy-federated.uc.r.appspot.com/
# 
[jacky] https://mozilla.social/@wedistribute/111660443230760980 this feels very non-Webby
# capjamesg I find the table on https://20231226t135650-dot-bridgy-federated.uc.r.appspot.com/ overwhelming.
# capjamesg There is too much information and I really need to focus to figure out the small differences between each item in the table.
# [snarfed] as discussed in https://snarfed.org/2023-03-24_49619-2
# Loqi [preview] [Ryan Barrett] Content negotiation considered harmful https://snarfed.org/matilda-used-car-salesman-harry-wormwood.jpg
# [snarfed] ugh I still can't interact with mozilla.social because of *another* conneg bug, Elk's broken Accept header parsing: https://github.com/elk-zone/elk/issues/2463
# [snarfed] [tantek] lmk if it would help for me to file an issue on https://github.com/MozillaSocial/elk too, or submit this to the feedback form https://survey.alchemer.com/s3/7470063/mozilla-social-feedback-form , or whatever
# [snarfed] [pfefferle] looks like this is another instance of https://github.com/Automattic/wordpress-activitypub/issues/580#issuecomment-1830443110 . I suspect this will hit many of the plugin's users. you might want to consider setting the Vary header yourself to fix it
# 
[pfefferle] [snarfed] I am not really sure what it might break on the other hand! It will vary for every Accept header and because browsers often use different ones, it might flood caches.
# [snarfed] [tantek] oh https://github.com/MozillaSocial/elk is a fork so it doesn't have issues. any idea where else you want it?