#dev 2020-06-24

2020-06-24 UTC
#
[chrisaldrich]
jmac, I always think of read, listens, watches (and probably a few others) as bookmarks with some additional semantic meaning. I think bookmarks are reasonably popular. I send lots of read posts and many sites interpret (read: display) them as basic bookmarks rather than simple "mentions".
[jgmac1106], wombelix and geoffo joined the channel
#
jmac
I gotta add mine to that list
#
jmac
Guess that’s my summit-week project
#
[tantek]
what is archive
#
Loqi
archive in the context of the indieweb refers to date-grouped (often monthly) sets of posts (AKA personal historical archives, a common form of navigation), but can sometimes mean archival copy, a copy of a web page made (often by someone other than the author) at a particular point in time https://indieweb.org/archive
#
[tantek]
archive << UI considerations, language and cultural conventions on time, past vs future etc.: 2017-06-13 [https://theconversation.com/language-alters-our-experience-of-time-76761 Language alters our experience of time]
#
Loqi
ok, I added "UI considerations, language and cultural conventions on time, past vs future etc.: 2017-06-13 [https://theconversation.com/language-alters-our-experience-of-time-76761 Language alters our experience of time]" to the "See Also" section of /archive https://indieweb.org/wiki/index.php?diff=70507&oldid=66858
[snarfed] joined the channel
#
jacky
been working to consider the MVP of my stuff the state at which I can use it for my personal site comfortably
#
jacky
that might be a double edged sword because for Lighthouse, I want to be able to send/receive Webmentions but I'd also need proper async Webmention support
#
jacky
everything else after that would be things for people who don't run Webmention servers
#
jacky
which brings to me something; do we have an idea around a 'standard' for Webmention-centric feeds? I'm leaning on having a 'list' of h-cites since that's the most natural MF2-y approach
#
[tantek]
do you mean like a /notifications page?
#
jacky
nah (but close!); it'd be the feed one would plug into their Microsub server so they can get some sense of notifications
#
jacky
synthetic-uris++
#
Loqi
synthetic-uris has 1 karma over the last year
#
jacky
I'm leaning into making these for some pages in Lighthouse (from sites that don't have MF2)
nickodd, maxwelljoslyn and justache joined the channel
#
[chrisaldrich]
jacky, this sounds like the sort of page gRegor has for plugging into microsub notifications: https://gregorlove.com/responses/. I wonder if there are other documented examples?
#
jacky
interesting yeah
#
gRegorLove
I subscribe to that in the Aperture notifications channel so it shows up in Monocle
#
jacky
that's the kind of feed I'd want for people
#
gRegorLove
those are h-cites as well
#
jacky
takes a note
#
jacky
I might want to see if I can protect this page for people (responses aren't necessarily meant for everyone)
#
jacky
but this is a dope start
#
gRegorLove
Yeah, my page only shows webmentions that are public on my site. I can manually hide a webmention, though rarely do.
#
gRegorLove
When I'm logged in I can see webmentions pending processing too
#
gRegorLove
Good case for Autoauth if you want to protect it :)
treora joined the channel
#
jacky
yup! I do need to look into implementing that
#
jacky
but I wanna externalize my Webmentions first (to do proper support of async Webmentions)
#
ben_thatmustbeme
loqi, got any messages for me?
#
Loqi
ben_thatmustbeme: tantek left you a message on 2020-02-12 at 6:25pm UTC: tomorrow is the 2 year anniversary of the closing of the W3C Social Web Working Group! https://www.w3.org/wiki/Socialwg if you have any particular updates regarding SocialWG/IndieWeb specifications that you worked on or implemented, consider this a good time to post about
#
Loqi
ben_thatmustbeme: jeremycherfas left you a message on 2020-05-22 at 8:20am UTC: I just got a somewhat lackadaisical reply from Samuel Clay to the continuing saga of trying to get Newsblur to parse h-feed. I guess you got it too. Is there any point in keeping the issue alive?
#
ben_thatmustbeme
hmmm, jeremycherfas, probalby not
justache, ketudb, KartikPrabhu, [fluffy] and [chrisaldrich] joined the channel; nickodd left the channel
#
[chrisaldrich]
Is anyone else having issues with [snarfed]'s https://instagram-atom.appspot.com/? It had been working recently, but the feed has died and I can't seem to resurrect it. Is Instagram playing games again?
swentel, swentie, [KevinMarks] and [tantek] joined the channel
#
[tantek]
welcome back ben_thatmustbeme++
#
Loqi
ben_thatmustbeme has 1 karma over the last year
dckc, moppy, superkuh, [LewisCowles], gRegorLove, gxt, [jgmac1106] and KartikPrabhu joined the channel
#
[jgmac1106]
kevinMarks I wonder if you should follow up your webmention on https://kevq.uk/how-to-create-an-indieweb-profile/ and point them to other ways you can link to an h-card without display: none
#
petermolnar
I found an interesting firefox bug(?): if images are set to html lazy-load, the print-preview doesn't load them until the page was scrolled to the bottom and all images loaded.
#
Loqi
swentel: jacky left you a message 3 days, 1 hour ago: it begins! https://github.com/swentel/indigenous-android/issues/379 I'll begin working on this _slowly_ because I've finished sketching out what my idea for a mobile IndieWeb experience would look like but I'm also using this as a chance to contribute :)
#
sknebel
thats an interesting one. kinda makes sense (not load more images than the user has seen when they pressed print), but could be confusing
#
sknebel
or is it not-showing *all* lazy-loaded images?
#
swentel
and does it print the images, even if you did not scroll?
swentel and gxt joined the channel
#
petermolnar
both are good questions, let me see
#
petermolnar
it does not print them
#
petermolnar
so that's one of those it's technically correct, but not how it's supposed to be behaviour
KartikPrabhu joined the channel
#
petermolnar
bug submitted under https://bugzilla.mozilla.org/show_bug.cgi?id=1648064 - I'm curious to see the outcome, because it's an indicator on what the normal/expected behaviour is these days.
[jgmac1106] and KartikPrabhu joined the channel
#
@_stefanzweifel
↩️ Not my content; but from others I respect. However, the stolen content has outgoing links to my site and through the power of webmentions I got a notification. I see stolen content from professional companies all the time at my job. Don't understand why someone copies blogs.
(twitter.com/_/status/1275776816695967746)
[schmarty] and [snarfed] joined the channel
#
[snarfed]
[chrisaldrich] sorry, it gets rate limited a lot 😢
geoffo, dckc and sivy_ joined the channel
#
sivy_
anyone know of a repo of html samples of microformats
#
sivy_
i find sometimes it’s hard to read the indieweb or microformats wikis and imagine what the expression in HTML looks like
#
aaronpk
there's the test suite which has a bunch https://github.com/microformats/tests
#
Loqi
[microformats] tests: Microformats test suite
#
sivy_
nice
#
sivy_
perfect
[CrowderSoup], hs0ucy and [tw2113] joined the channel
#
sivy_
I’m working on my webmention implementation, and post-type discovery is confusing
#
sivy_
having to dive into the content to find the source link, then back out to the h-entry, then … <where> … to figure out the post type
#
sivy_
the h-entry is on the div
#
sivy_
and the in-reploy-to is on the div->div
#
sivy_
:facepalm:
#
aaronpk
the way you're talking about this sounds like you're not traversing the parsed microformats tree
#
aaronpk
you shouldn't be worried about divs, you should just be looking at the parsed result like this http://pin13.net/mf2/?id=20200624150850395
[jgmac1106] joined the channel
#
sivy_
@aaronpk I’ll hjave to do some more tests, posible that the parser i’m using has some limitations
#
sivy_
tanks
#
sivy_
—-(==)
#
sivy_
(======)
[chrisaldrich] joined the channel
#
sivy_
@aaronpk how exactly do I answer this? “if its hyperlink to the original post has an in-reply-to url”
#
sivy_
seems like you have to determine in-reply-to before you get a url
#
sknebel
you run a parser over it and search for the url in its output?
#
sivy_
well, the issue is that the link doesn’t have a in-reply-to URL, the in-reply-to h-entry has a
[schmarty] joined the channel
#
[schmarty]
sivy_: sounds like it's marked up as "u-in-reply-to h-entry"
#
[schmarty]
if that is the case, your parser should generate, for the "in-reply-to" attribute, a nested h-entry but *also* provide a "value" attribute that contains the URL you are looking for. the "u-" in "u-in-reply-to" means that parsers should look for a URL.
#
sivy_
have very different values for hentry->in-reploy-to
#
sivy_
there’s a LOT of variation in the MF “object model” :)
#
sivy_
no surprise to those who have been here a while but makes implementation… interesting
#
[schmarty]
haha, there can be! but there are some safety valves for limiting how much complexity you take on 😄
#
sivy_
specifically I’m trying to determine the type of a webmention, but the model doesn’t really support
#
sivy_
“the <a> for link X has property Y
#
[schmarty]
in this case both parsers have an in-reply-to property that's an array. in the second example, the only value in that array is a string (presumably the URL, but it should be validated). in the first example, the only value in that array is an object with type, properties, and value. if you want to dive in to type and properties you can, but "value" is the equivalent of that string.
#
sivy_
it’s “the hentry X has properies Yn, any of which might be the link to the mentioned post”
#
[schmarty]
sivy_: ah, yes, that is true!
#
[schmarty]
and a given post may "be" a reply or mention or etc for multiple other posts.
#
sivy_
URLS here are commonly the inner-most element
#
sivy_
yeah
#
[schmarty]
those "properties Yn" of "h-entry X" are what the Post Type Discovery algorithm is referring to https://indieweb.org/post-type-discovery
#
[schmarty]
(sorry if this is redundant. trying to make sure i understand the issue you're running into.)
#
sivy_
yeah, I’ve gotten that now; means I have to check each type and then look for a url that matches the target
#
sivy_
annoying but I think i’ve got the pattern worked out
#
sivy_
(also makes me wish I was working in python for this bit instead of golang — when a value could be a string or map staticly-typed langs are annoying)
deathrow1 and [snarfed] joined the channel
#
[snarfed]
sivy_ you probably want to decouple webmention verification from post type discovery. the former is easier with the raw HTML. (i literally just do a string search across the whole doc.) for the latter, you follow the PTD algorithm against the parsed mf2. very different tasks
#
sivy_
I am verifying the webmention via string contains
#
sivy_
but when storing the webmention I need the post type
#
sivy_
so I’ve already determined that the hentry contains my link
#
sivy_
but need to determine the type of reference
#
sivy_
and theoretically at least, an hentry could include a comment on one link and a like on another
#
sivy_
though authoring such would be a feat
gRegorLove joined the channel
#
[snarfed]
but yeah the common case is single
#
[snarfed]
so maybe just handle that and leave the multi case as a todo. could check for multiple different links to your site in the verification step and alert yourself the first time it happens. might not be for years
KartikPrabhu and garrettw joined the channel
#
sivy_
[snarfed]: ayup
#
sivy_
[snarfed]: though the MF parser returns ‘bout everything as lists anyway, so already looping over them, doesn’t hurt to go ahead an check those multiple values
geoffo joined the channel
#
[snarfed]
sivy_ oh yeah it's not the code as much as the semantics. like, if you get a u-edit-of, or u-location-of, or something else exotic, will you know what to do with it?
#
[snarfed]
that's where PTD and only looking at specific properties helps, so that you only specifically handle types you're actually ready for. and then otherwise interpret as a generic mention
#
sivy_
yeah. messy :D
#
jacky
I had some thoughts about async Webmentions which brought me back to making a Webmention endpoint a bit more 'powerful' https://v2.jacky.wtf/post/57fbda1a-d12a-474e-aed0-0811885f6b0a
#
Loqi
[Jacky Alciné] I wrote up some thoughts on how I’ll have Koype handle Webmention configuration remotely over its IndieWeb wiki page. Right now, it’s a bit theortical but this will work well since it’s based on the work of the Micropub specification. I almost ...
#
jacky
I'm inching towards more things that enable "one click"-y setups for people
#
sivy_
hi jacky
#
jacky
what's up sivy_
#
sivy_
oh you know MF parsing and webmentions
#
sivy_
slowly getting my head around the post-type algorythn
#
jacky
yeah if it's not in PHP or JavaScript, you're writing it up from scratch
#
jacky
you can cheat a bit by wrapping calls to a service like xray
#
[schmarty]
jacky: i read through the proposal but i got confused about the configuration bit
#
jacky
yeah so let's say you signed into webmention.io
#
[schmarty]
so if i went through the get-a-token flow with the "webmention" scope and then made a micropub q=config query
#
[schmarty]
i'd see a webmention_configuration_endpoint
#
jacky
ah yeah ofc
#
[schmarty]
seems like i can then use that token to make requests to the URL in webmention_configuration_endpoint. the actual shape of those requests is TBD but includes getting/setting things like the webmention endpoint and other WM-related things like vouch lists, etc.
[chrisaldrich] joined the channel
#
[chrisaldrich]
thanks snarfed re: instagram appspot tool. I eventually saw a rate limit response somewhere else as feed readers weren't helpful. I figured Instagram was the bad actor, but was partially worried that it was a PEBKAC problem.
#
[schmarty]
i think i get it - the thing i was missing is that webmention_configuration_endpoint probably has direct access to my CMS (and more) so it can update those values
[CrowderSoup] joined the channel
#
[schmarty]
(e.g. calling webmention_configuration_endpoint to set a new webmention endpoint would make that the new value for <link rel="webmention" ...> on posts on my site?)
#
jacky
[schmarty]: yup!
#
[schmarty]
got it! and very interesting.
#
jacky
I wanted to give a chance to make a separate endpoint for those who don't have full control over their Webmention endpoint (users of Webmention.io from a static site for example)
#
Loqi
jacky has 28 karma in this channel over the last year (96 in all channels)
#
jacky
I want to make things into buttons! lol
#
[schmarty]
i've been thinking about how to make a minimal-but-complete indieweb playground and wanting to automate a bunch of the "plugging together" as well
#
[schmarty]
part of that is easy if i "just" build most of it as a monolith
[pfefferle] joined the channel
#
[schmarty]
but i've got this separation in mind where the public site is hosted statically (e.g. with neocities) and files there are updated and added via an app that is basically translating the indieweb building blocks into minimal changed
#
[schmarty]
*changes to your statically hosted files
#
jacky
yeah something like this would take time to crunch for a static site (especially if it has like tens of thousands of posts)
#
[schmarty]
hahaha yeah changing that webmention header has to touch every post that accepts WMs on a static site 😂
#
Loqi
nice
#
jacky
this is obviously only going to be on my site for now
#
jacky
but I'm 100% going to be working on a series of 'guides' (trying to work with someone on this) for like a $POPULAR_FRAMEWORK to get it from 0 to indieweb-adjacent and use that to 'ship' these features
#
[schmarty]
looking forward to that!
[KevinMarks] and [grantcodes] joined the channel
#
[grantcodes]
what is indiekit?
#
[grantcodes]
Sounds like that schmarty
#
[grantcodes]
Well a little
#
[schmarty]
thanks for the reminder, grantcodes! i think micropub endpoints built around supporting static sites (of which i am guilty of building at least a couple) are similar in many ways.
#
[schmarty]
my hope is to start figuring out a minimal way of replacing the static site generator build step (build all the files every time) with something that does as few transformations as possible.
#
[grantcodes]
I'm also starting to use Postr in a similar way. Basically it becomes a api + storage (db + media in my case). And handles micropub and webmentions and all that. But my frontend is static / serverless
#
[grantcodes]
What are incremental builds?
#
Loqi
It looks like we don't have a page for "incremental builds" yet. Would you like to create it? (Or just say "incremental builds is ____", a sentence describing the term)
#
[grantcodes]
Hmm I think that is what you're looking for in the ssg world
#
jacky
wants the term serverless to die, lol
#
[schmarty]
grantcodes: i've been waiting for incremental build support in hugo for so long that i have ceased to hope, lol. i think gatsby added it recently?
#
petermolnar
joins jacky with a torch and a pitchfork
#
[grantcodes]
I think new jekyll and gatsby support that and next.js and eleventy in beta / on the way
#
[schmarty]
i have waited long enough that i have soured a bit on SSGs and i want to try another approach.
#
[grantcodes]
Heh it's hard to do dynamic + static
#
[schmarty]
i'd like to try and skip the "pile of source files that a compiler turns into a website" and instead just operate on the website
#
[schmarty]
lots of work to go before i have something worth talking about here, probably.
#
[schmarty]
maybe a good discussion for the weekend tho!
#
[grantcodes]
Good luck with that! 😅 if you have a pile of content you have a pile of files somewhere
#
[schmarty]
oh for sure - but i'd like to try "a pile of files that is already a website"
#
vilhalmer
html is a good format
#
Loqi
html has 1 karma in this channel over the last year (2 in all channels)
#
vilhalmer
also makes it really easy to test, just fire up caddy's file-server in the directory and you're done
garrettw, [schmarty]1, [KevinMarks]1 and [CrowderSoup]1 joined the channel
#
Loqi
Lighthouse
#
jacky
(this is on my laptop so be gentle lol)
#
jacky
okay nice; that's working correctly and patched a bug that'll be effectively _everywhere_ in my apps lol
KartikPrabhu joined the channel
#
petermolnar
I don't have a better place to discuss this: https://bugzilla.mozilla.org/show_bug.cgi?id=1648064 (loading="lazy" images are not printed unless the were scrolled to)
#
petermolnar
I found this bug this morning, and my questions is the following: which would you consider the expected behaviour, images you have not yet scrolled to appear on the print or they don't?
[schmarty] and [chrisaldrich] joined the channel
#
aaronpk
i would definitely expect them to print
#
aaronpk
the act of choosing to print the page should cause the rest of the images to be downloaded
[KevinMarks] joined the channel
#
[KevinMarks]
How does chrome behave in that case? I agree with Aaron that it should download the images for all the pages that are printed.
#
jacky
does it matter how Chrome behave? (has a penchant against explicit measurement of browsers against each other)
#
[KevinMarks]
I meant that you may need to file the same bug against chrome
#
petermolnar
[KevinMarks]: the same; Chrome it doesn't want to print them either
#
jacky
sounds like potentially something that should be spec-filed?
#
petermolnar
the tricky part of the question is that every lazy-loading JS behaves like this, as in not printing unless scrolled to
#
petermolnar
and yes, I need to file the same bug in chrome... sigh. Where do I do that?
#
aaronpk
i think it's fuzzier with JS, the answer is pretty clear with the img tag tho
#
aaronpk
it's lazy-loading, not not-loading
#
jacky
petermolnar: I think http://crbug.com
#
jacky
or crbug.org
#
aaronpk
where lazy is "not loaded until needed", where "needed" is either when scrolled to or when printed
#
jacky
I can't remember
#
[KevinMarks]
What happens when you print an infinite scroll?
#
aaronpk
heh nice and vague "The attribute directs the user agent to ... defer fetching until some conditions associated with the element are met"
#
aaronpk
[KevinMarks]: that's why i said the answer is fuzzier with print
#
aaronpk
with JS
#
petermolnar
[KevinMarks]: I'm going to sell old dot matrix printers with those boxes of perforated edge papers under the label of "print your whole facebook wall or twitter feed!"
cjw6k joined the channel
#
[KevinMarks]
Is svgshare.com a good test case?
#
[KevinMarks]
They're lazy loaded
KartikPrabhu joined the channel