#dev 2023-06-20

2023-06-20 UTC
[benatwork] joined the channel
#
[benatwork]
Oh weird. Thanks for the heads up
#
[tantek]
Also it looks like RSVPs on your event posts aren't showing up, e.g. the event that RSVP is in reply to: https://werd.io/event/51c94deebed7de5c2386a1b4/unofficial-portland-post-indiewebcamp-meetup
#
Loqi
[preview] [Tantek Çelik] @renegadejade@hachyderm.io in short, yes, it is a thing! Lots of great replies to your post already about directly using #WithKnown or #WordPress with the #ActivityPub plugin, or cross-posting from/via GitHub. Another #IndieWeb compatible option i...
eitilt, bterry1 and ahappydeath joined the channel
#
[snarfed]
tantek++ thanks!
#
Loqi
tantek has 29 karma in this channel over the last year (101 in all channels)
alephalpha0, tei_ and [timothy_chambe] joined the channel
#
[timothy_chambe]
Over inside the Fediverse, lots of chatter this weeknend on Meta/instagram's "Project 92" and much back and forth sussing out the threat of an "Embrace, Extend, Extinguish" play they might do against ActivityPub.
#
[timothy_chambe]
Some questions:
#
[timothy_chambe]
Maybe the podcast format vs Spotify or Apple extensions?
#
[timothy_chambe]
One - if anyone here has any thoughts on the validity of that specific concern... If people here have any ideas on how to combat an "EEE" strategy in that case from Meta.
#
[timothy_chambe]
Two: if people here have examples of past times a company tried to do a "EEE" effort against an open spec that was successfully thwarted, and what strategies and tactics proved to work.
#
[timothy_chambe]
Keeping React Framework interoperable even though it was born out of FB?
#
[timothy_chambe]
Others?
[aciccarello], jeremycherfas, wagle, gerben, IWSlackGateway, aaronpk, nertzy, tei_ and [tantek] joined the channel
#
[tantek]
[timothy_chambe] EEE is a concern, and especially so because there is no federation test suite for ActivityPub (e.g. an equivalent of webmention.rocks)
tei_1 joined the channel
#
[tantek]
Absence of a test suite makes a spec vulnerable to monoculture implementations or ecosystems, which is why Mastodon as the current dominant implementation can innovate and other AP implementations have to "be like Mastodon". Obviously if 10-100x users start using something else (whether Project 92 or something else), then that something else would have that ability
[manton] joined the channel
#
[manton]
My 2 cents, I agree EEE is a concern but having big platforms support open platform is what we should want! If Project 92 has an even halfway decent ActivityPub implementation, that’s way better than a closed silo with no API or a private API.
#
[manton]
I don’t get the suggestion (elsewhere) that instances should de-federate with Meta before we even give them a chance to do the right thing.
#
[tantek]
[manton] agreed, I think we should presume good faith on the part of this team until proven otherwise, and perhaps even encourage them to help contribute to an ActivityPub server-to-server open test suite as part of their development
#
[manton]
One way I could see it playing out is that the ActivityPub server support is fine, but there’s no client API. No C2S or Mastodon API or Micropub or anything, so you have to use their own app. Not terrible but not ideal either.
#
[manton]
That would be the most consistent with Meta’s other platforms and having an ad-based business where they want to control the UX.
#
[tantek]
[manton] indeed. It's ok for them to have their own app, we should request that their app support the Micropub client side and their server support the Micropub server side, since there are clear test suites for both
#
[tantek]
AP C2S really hasn't gotten much if any adoption and it's had a long enough chance (same W3C REC as AP S2S) that I don't see any reason to dwell on it or encourage it. If people want to experiment sure, but there's no ecosystem there. In contrast there is a Micropub open ecosystem of clients & servers, and there is the Mastodon-proprietary client API which a number of clients have adopted (but no other servers AFAIK)
#
vladimyr
Mastodon API is available on Pleroma/Akkoma, Calckey (not sure about misskey tho) and friendica iirc
#
vladimyr
It became defacto standard c2s api
[schmarty] joined the channel
#
Loqi
ok, I added "https://www.zachleat.com/web/eleventy-side-project/" to the "See Also" section of /Eleventy https://indieweb.org/wiki/index.php?diff=88327&oldid=87310
#
c​apjamesg
😢
#
[tantek]
vladimyr++ thanks for the update, appreciated. Is there a test suite for the Mastodon API or is everyone on a hamsterwheel keeping up with Mastodon PRs?
#
Loqi
vladimyr has 4 karma over the last year
tei_ joined the channel
#
vladimyr
[tantek]: I'm not aware of such thing so I guess it goes as you said - playing catch up game with mastodon
[KevinMarks] joined the channel
#
Loqi
[preview] Thoughts from “Meet Safari for Spatial Computing”
#
[KevinMarks]
I had an interesting CSS time for dayjob today - I had got too used to being on my giant iMac screen and hadn't been cross testing on smaller displays, so the loose form layout I was using was pushing all the content offscreen. I came up with a test model of having iframes set to the laptop viewport of some client users' screens to crosscheck scaling. 1280x608 was a lot less vertical space than I assumed.
[Murray] joined the channel
#
[Murray]
Polypane is great for that kind of thing: https://polypane.app/ (paid though)
#
[tantek]
I mean, a bunch of browser (including FF) have one-menu-item-click responsive design mode for checking this sort of thing
#
[tantek]
Tools > Browser Tools > Responsive Design Mode
#
[tantek]
cmd-option-m to quick toggle on/off — works great
#
[Murray]
Sure, I think all Western browsers have that? Not as useful as something like Polypane when developing, but incredibly useful if you have nothing else 🙂 (I actually couldn't imagine doing my job without that feature, and FF's is my favourite by far, mainly because you can use it independently of the code inspection panel)
#
[tantek]
thanks [Murray] — it's literally been one of the most useful "check in browser" features I've used too
#
[KevinMarks]
In this case it isn't meant for mobile at all, so I was doing a desktop/laptop responsive grid, but had forgotten how vertically squished some laptops are. I just tried FF with a custom version set to that 1280x608 as the default 1280x800 is wrong (It is 1920x1020 scaled to 1280x680, but I lose a bunch to OS and browser chrome, which the FF thing doesn't allow for)
#
[tantek]
vertically squished laptop can be approximated by landscape mode mobile 🙂
#
[0x3b0b]
I used to get pretty unreasonably angry at things that seemed to assume I'd view them at a width of 1280. I've mellowed a tiny bit on that topic since I stopped using 1280x1024 screens and also stopped having much to do with the $dayjob interface that I found to be the worst offender.
#
[KevinMarks]
That's fair; in this case it's a pretty data dense dashboard that is meant to be an overview of complex stuff, so the design centre was bigger screens, so even if it was an odd shape it was 1920x1080 so lots of space for the form controls.
#
[tantek]
hah indeed classic 1080p
#
[0x3b0b]
Yeah, that makes sense. $dayjob also has at least one internal application that seems to have been designed specifically on the assumption that the staff using it would have dual monitors with one of them portrait. Since it was internal and for on-site staff, that's an assumption the company could ensure was provided for.
#
[KevinMarks]
In this case it started out with dinky laptops like that ~5 years ago when I was first working on it, but since 2020 most of the users are now using bigger external screens, so space creep wasn't so obvious as I added things, but one of the team was using it on their laptop somewhere.
#
[0x3b0b]
The site that bugged me had the logout link - and I think possibly one other sometimes-important piece of UI - where it wound up requiring me to scroll to the right to click on it because the ~100px width of my vertical taskbar wasn't allowed for. And what really bugged me about it was that it was only that far over because of the width of a header image that was mostly a repeating pattern. I think we fixed it eventually.
#
[KevinMarks]
the other thing I realised is that I forgot that windows machines have visible scrollbars on overflow as I'm so used to MacOS hiding them, and that exacerbates things
#
[0x3b0b]
If I'm not mistaken that's a default rather than a guarantee, and possibly also browser-specific; but whenever I come across something with auto-hiding or auto-shrinking scrollbars that I wind up using regularly, I go looking for how to prevent that; so I'm not sure.
#
[tantek]
scrolltotheright--
#
Loqi
scrolltotheright has -1 karma over the last year
#
[KevinMarks]
quite - I had accidentally got right scrollbars on 2 columns of the layout in that screensize, as well as the vertical ones I expected
#
[KevinMarks]
(the workaround was zoom out with ctrl minus, but less than ideal)
#
[KevinMarks]
also the FF screen tool doesn't behave right with zoom - it should zoom the scaling inside the cropped window, but it makes it smaller instead.
#
[0x3b0b]
Same result in Vivaldi. To be fair, I can see that also being a useful thing to be able to do, but...
#
vladimyr
As 0x3b0b pointed out having overlay scrollbars hidden on mac is indeed default but as soon as you plug in mouse that reverts to always visible
#
vladimyr
And it can be further changed and configured through system prefs
#
vladimyr
What can help you is reserving space for scrollbars with scrollbar-gutter: stable https://developer.mozilla.org/en-US/docs/Web/CSS/scrollbar-gutter
Wildernetic, tei_, lockywolf, gRegor, IWSlackGateway, [tantek], [KevinMarks], epoch_, gwg_, sivoais and [timothy_chambe] joined the channel
#
[timothy_chambe]
[tantek] and [manton] - great thoughts and notes. I'll check to see if any test suites are in the works.
#
[timothy_chambe]
On the other question can anyone here think of good examples where a EEE campaign against an open standard was thwarted well? And what common traits that could be lessons learned from those that didn't succum to being "embraced, extended and extinguished" that folks can learn from?
#
[timothy_chambe]
Having a robust test suite for one tactic is definitely one.
#
[timothy_chambe]
To me Podcast formats seemed resilient to Spotify or Apple trying to extend it. I heard others say that even though React framework came out of FB, it opened up broadly and became FB independent well. Others?
jjuran joined the channel
#
vladimyr
React is anything but good example cause it simply changed hands, from FB to Vercel. Good luck trying to push something that doesn't align with their NextJS supporting library agenda
#
[tantek]
[timothy_chambe] the assessment of Podcast formats is incorrect. Due to neglect/failure of any form of RSS standardization in an actual standards org, Apple was able to easily extend "podcast feed format" with tons of proprietary extensions that anyone who wants to actually have a podcast work in tools etc has to use
#
[tantek]
I don't know about Spotify impacting or not any podcast formats
#
[tantek]
to be fair to Apple, many (most?) of their podcast extensions had good use-cases. only some of the extensions were redundant with what you could already do/infer with RSS
#
[tantek]
and there's been no evidence of any attempt to "extinguish" RSS-based podcast formats that I know of.
#
[tantek]
more the first part of EE happening rather than full EEE
#
vladimyr
> What I learned is that Apple does not produce an RSS feed for podcasts that they host. That’s right: if you host your show with Apple, the only listeners you can have are folks with the Apple Podcasts app.
[snarfed] joined the channel
#
[snarfed]
they even make interop for _inbound_ RSS feeds difficult, eg https://github.com/snarfed/granary/issues/221
#
Loqi
[preview] [snarfed] #221 RSS feeds don't work with iTunes because Apple doesn't accept Google's SSL certs
#
[timothy_chambe]
Thanks [tantek] on the latest on the Apple's effort to "extend" podcasts. I might look to see if that "worked" did podcast producers start to use that, or did the community retain interoperablity.
#
[timothy_chambe]
[tantek] and [manton] - on the topic of some instances (about 3 percent) signing a pledge to defederate Meta premptively. None of the big servers have, and almost none of the mid-range ones. Indieweb.social is at mid-range with about 11,000 users, and here was my statement on our policy: (open for any suggestions or notes) https://indieweb.social/@tchambers/110568723404841278
#
Loqi
[preview] [Tim Chambers] ➡️ #Admin #Indiewebsocial With news of the probable launch of Meta's #Project92 I wanted to make clear this servers policy:"Don't preemptively strike meta w/ a fediblock, but stay vigilant with eyes wide open and a finger on the block button." ...
#
[tantek]
good find vladimyr. that is indeed quite concerning and sufficient reason to not "host" a podcast "with Apple" (whatever that means, I always thought hosting postcasts was something people did on their own domains?)
#
vladimyr
They might host rss feed on their own domain, but typically audio files storage is located at different corner of the internet
tei_ and [schmarty] joined the channel
#
[schmarty]
Apple offers a premium podcast service where you can host your podcast, get a ton of analytics about your listeners, accept paid subscriptions, and more. No RSS if provided for these and they are only available through Apple's app.
#
vladimyr
And audio is tainted with DRM so unless you keep your audio sources safe you can't really backup podcast episodes
#
[tantek]
hmm, income/analytics streams in exchange for proprietary exclusivity.
#
jjuran
Faustian bargain
#
[timothy_chambe]
[tantek] or others: what would it take to get even a basic "alpha" version of a ActivityPub test suite up and running in the next weeks assuming a Barcelona launch is immenant and getting one live asap was key. Would that need to be done formally inside any standards org or would develpers outside starting one still be productive?
#
[tantek]
[timothy_chambe] anyone could "start" one, and frankly, anyone could have since 2018, but it's a pretty big red flag that NO ONE across the "ActivityPub community/s" (and I mean that in quotes) has bothered to for over five years should tell you something.
#
[tantek]
thus good luck on getting anything AP S2S up & running in "weeks"
#
[tantek]
I have repeatedly asked *multiple* AP devs to start/contribute to an S2S test suite and crickets / tumbleweeds
#
[schmarty]
cwebber's initial test suite is open and i believe she has pushed several times for someone to take over. https://socialhub.activitypub.rocks/t/the-activitypub-test-suite/290
#
[tantek]
I have pointed to webmention.rocks as an example so they don't have to reinvent the wheel of how to start a test suite site
#
[tantek]
[schmarty] no that's not helpful. cwebber's test suite was C2S only — the half of AP that has failed to get any serious uptake
#
[schmarty]
ah, yes, that is definitely an issue
#
[KevinMarks]
I think cwebber's suite relies on some unusual tooling that may be hard to get running again.
#
[tantek]
yes I've seen that post [timothy_chambe]. It's not really helpful to S2S and frankly mischaracterizes the S2S test suite problem/challenge (because the webmention.rocks test suite existence, functionality etc. disproves a lot of cwebber's assertions about S2S test suites).
#
[KevinMarks]
If you have an async protocol, an async test suite is necessary
#
[tantek]
handwave handwave handwave
#
[tantek]
IMO the excuses for "why it can't be done" are nonsense. the excuses for "but it's HAAAARD" I have some sympathy for, except if you're putting yourself forth as an "expert", grow a backbone and step up to take on and solve the hard problems, otherwise stop putting yourself on a pedestal as an expert
#
[timothy_chambe]
Anything in this fork seem promising? (but seems dead and not updated in years) - just looking to see if there are any efforts to build on: https://github.com/go-fed/testsuite
#
Loqi
[preview] [go-fed] testsuite: Unofficial ActivityPub test suite server, which runs machine-assisted test cases against federated software.
#
[schmarty]
[timothy_chambe] snarfed and manton have been doing great work bridging posts into and out of ActivityPub / Mastodon, but there aren't many developers dedicated to ActivityPub projects that hang out in this chat.
#
[tantek]
and it has been non-trivial to get things working like say, hashtags, across implementations
#
[tantek]
so the absence of a test suite that tests how those things are actually supposed to interoperate has been a real problem (and caused a lot more time to be spent reverse-engineering than should be necessary for a standard, especially one with examples)
#
epoch_
I'm not dedicated to ap, but I've written my own activity pub server software
ahappydeath joined the channel
#
[tantek]
Nice. That's quite an accomplishment! Can you add it to (if it's not there already) https://indieweb.org/ActivityPub#Software ?
#
epoch_
uhhh, sure. it is a bit of a mess and I kind of went off-spec for parts of it though.
#
[tantek]
even more reason! both of those are good sources of information about the spec and/or reasons to improve it
#
epoch_
checks how out of date the readme is
gxt__ joined the channel
#
[tantek]
New W3C website (mostly design, some content) dropped: https://www.w3.org/
#
[tantek]
speaking of ActivityPub, I see evanpro has been documenting a bunch here: https://www.w3.org/wiki/ActivityPub/Primer
#
[tantek]
ActivityPub << Developer primer (in progress) on W3C wiki: https://www.w3.org/wiki/ActivityPub/Primer
#
Loqi
ok, I added "Developer primer (in progress) on W3C wiki: https://www.w3.org/wiki/ActivityPub/Primer" to the "See Also" section of /ActivityPub https://indieweb.org/wiki/index.php?diff=88329&oldid=88328
holiday_medley joined the channel
#
[tantek]
what is a static site
#
Loqi
A static site is a website that serves pages directly from files on the web server, which have often been previously generated by static site generator software only when created or updated, instead of for every request https://indieweb.org/static_site
#
[tantek]
hmm that page really doesn't have much best practices advice on it
#
[tantek]
anyone have opinions on static site best practices? like a site that literally most of the content is / will be static, infrequently updated, certainly nothing like a stream of posts
#
[tantek]
like a very slowly/minimally edited wiki I suppose
#
[tantek]
certainly not worth the effort of installing / updating / maintaining an SSG
#
[tantek]
but I do still care about "clean" URLs, e.g. pathnames of words / phrases that are easily readable, well indexed, and no extra cruft like filename extensions like .html
#
[tantek]
added ^^ to https://indieweb.org/static_site#Brainstorming if anyone wants to comment, expand, provide other thoughts