#dev 2023-06-05

2023-06-05 UTC
ahappydeath and [snarfed] joined the channel
#
[snarfed]
this is plumbing error debugging, not UX
#
[snarfed]
and across two independent systems, ie whether a single protocol-level HTTP request does its processing synchronously, so that the client even gets an error at all, or not
#
[tantek]
even developers need UX. GitHub's existence is a good demonstration of that (over git CLI pain)
#
[snarfed]
definitely! this isn't even developer tooling though, it's code running in production
#
[tantek]
so devops UX then?
#
vladimyr
Used to be, with their recent move to new frontend that broke even the CTRL-F workflow it is really though sell
#
[snarfed]
there's no human user involved
#
[tantek]
like dashboards / cPanel etc.?
#
[snarfed]
this is during eg AP inbox delivery
geoffo joined the channel
#
[snarfed]
anyway. not a big deal, just wanted to surface a tradeoff of the common "do all nontrivial processing async" advice that we give (eg in the webmention spec) along with everyone else
#
[tantek]
if only there was an AP fed test suite
#
[snarfed]
would be nice! wouldn't really help here though, everything happening here is fully spec compliant
#
[tantek]
that could test both sync/async handling
#
[tantek]
a test suite can have error tests as well
#
[tantek]
good test suites do
#
[snarfed]
right, but again, all of those tests would pass here
#
[snarfed]
did you read the post?
#
[tantek]
looked like it was part of Bridgy docs? if I saw the link correctly
#
[tantek]
we need a webmention.rocks test for a "per-request status URL in the Location response header"
#
[tantek]
to encourage it to be more "widely implemented"
#
[tantek]
perhaps generalizing the "status URL in the Location response header" mechanism and making it a building block could help more protocols beyond Webmention
#
[snarfed]
definitely! sadly implementing it generally take a substantial amount of work, and primarily benefits senders, not the receiver who's implementing it. still really worthwhile overall though!
#
[tantek]
would be an interesting test if ActivityPub folks would be ok with depending on an external spec like that for extending the protocol
#
[tantek]
the benefit to the receiver *should* be that more senders are incentivized to implement sending to that receiver, because presumably that's the whole (higher level) purpose of putting up a receiver in the first place
#
[snarfed]
indirectly yes
#
[tantek]
giving senders the opportunity to provide better UX to their users provides incentive for those senders to work with your receiver
#
[tantek]
async successful task completion might need to be added to the "hard problems in CS" list
#
[tantek]
this goes way beyond web protocols
#
[snarfed]
lol yes, or at least status propagation
#
[snarfed]
arguably already there, async is just a temporal layer of indirection
#
[KevinMarks]
This is the structural complexity. The Promise model captures it, but the downstream paths can be hard to express in linear code. The async keyword model gives a clear path when things go right but masks the complexity of the failure cases
#
[snarfed]
right! and this adds another layer of difficulty in that errors are incorrectly reported as successes to clients, in order to benefit downstream server architecture
#
vladimyr
another /victim/ of async/await model is concurrency
#
vladimyr
I ment parallelism, sorry
#
vladimyr
allows you to express only serial but not parallel workflows
#
vladimyr
for that one you need to go back to Promises land
Soni joined the channel
#
[tantek]
In case anyone has a C64 laying around and wants to self-host from home: https://hackaday.com/2023/06/01/commodore-64-web-server-brings-8-bit-into-the-future/
Loqi__, sebbu2, gxt__ and holiday_medley joined the channel
#
Gorro_Rojo[theyi
gmorninn, does anyone know if there are some web standards for ecommerce sites? im thinking of something not unlike microformats, specially for small coop businesses
#
[tantek]
what is the small coop business use-case you are trying to solve?
#
[tantek]
it looks like IFTTT is still able to auto-crosspost to Twitter — is anyone using IFTTT to do their POSSEing? E.g.: https://twitter.com/w3c/status/1665671132681306113
#
[snarfed]
the new twitter API free tier allows posting, just not reading
[schmarty] joined the channel
#
Gorro_Rojo[theyi
<[tantek]> "what is the small coop business..." <- aaa darn I *just* found `h-product`, it's what I was looking for, sorry for bothering, my memory troubles betray me 😅 thank you!
#
Gorro_Rojo[theyi
<[snarfed]> "the new twitter API free tier..." <- that doesn't sound too bad
[timothy_chambe] and [tw2113_Slack_] joined the channel
#
Gorro_Rojo[theyi
it limits making third-party clients, but at least we can still posse to twitter, then
geoffo joined the channel
#
[snarfed]
that's a low bar, considering everything that _used_ to be free (or reasonably priced) and accessible in the Twitter API. https://snarfed.org/2023-04-03_so-long-twitter-api-and-thanks-for-all-the-fish
IWDiscord, moose333, [tw2113_Slack_], [KevinMarks], [fluffy]1, [timothy_chambe], [schmarty], [tantek], [snarfed] and IWSlackGateway joined the channel
#
[snarfed]
interesting, Apple just announced some kind of standalone web app support for arbitrary sites, including dedicated windows, dock icon, notifs. wonder if it uses any PWA standards under the hood or if it's entirely proprietary to Safari
#
shreyanjain[m]
I hope it's pwa
gRegor and [jgarber] joined the channel
#
Loqi
[preview] [Jen Simmons] Yes, you heard that right. Web apps on Mac, from Safari. *Any* website can be added to your Dock, and opened as a web app. Web app developers can customize their web app with a web manifest file and more — just like for web apps on iOS and iPadOS. ...
aaronpk joined the channel
#
shreyanjain[m]
Ahh that's nice, I don't use Safari but that's definitely nice to see better web app support right from Apple
#
[snarfed]
I dunno, it's Apple, their track record isn't exactly standards-first. we'll see
#
c​apjamesg
Yeah.
#
c​apjamesg
They said developers don't need to do anything.
#
c​apjamesg
Which makes me wonder.
#
shreyanjain[m]
That does not bode well 🤔
#
[KevinMarks]
Is this Web Clips again? Or whatever the widgets thing they had before?
#
[snarfed]
more like Electron evidently
#
[jgarber]
That sound you heard was my bones turning to dust.
bret joined the channel
#
[tantek]
The trajectory of their track record has been quite standards-strong for a few years now. I'm optimistic.
#
[tantek]
^ that was re: [snarfed] "track record" comment
[jacky] and geoffo joined the channel
#
[snarfed]
maybe web standards in Safari specifically? less so in other areas. still, that's good to know, and promising!
#
[tantek]
or in WebKit specifically, and this seems like an interesting expansion of the use of WebKit in the OS
geoffo, ahappydeath, 040AAT3TZ and gxt__ joined the channel
#
vladimyr
Oh there we go, Apple as we all know it: https://twitter.com/jensimmons/status/1665782319796346881
#
vladimyr
> Yes. “PWA” is a very specific term that requires both a manifest file and a service worker. Neither is required for web apps on Mac.
#
vladimyr
web standard = very specific term ✅
#
[tantek]
Jen is correct. Neither manifests nor service workers are necessary plumbing for what a user perceives as a "web app".
#
vladimyr
Definitely true from end user perspective but arguing with established standards doesn't really do users a favor cause now they need to go back to devs and ask them to support both pwa and webkit webapp
#
vladimyr
And then there is this also:
#
[tantek]
PWA is not an "established standard". There is no W3C spec for "PWA". PWA is dev-targeted marketing. 😛
#
vladimyr
Um, sure bout that? https://www.w3.org/TR/appmanifest/
#
[tantek]
yes I am sure. that link has zero results for "PWA".
#
vladimyr
Which is ofc marketing and umbrella term for manifest plus worker
#
vladimyr
My point wasn't in arguing with nonexistent pwa standard but manifest part that's supposedly redundant
#
[tantek]
webapp manifest is a standards-track working draft yes. there are also standard alternatives to many (most?) of the features contained therein.
bterry joined the channel
#
[tantek]
in other words, not requiring manifest does not mean you are not standards compliant or non-standard.
ahappydeath joined the channel
#
Soni
idea: there should be a lightweight fedi scraper you can run without having to setup a whole postgresql and whatnot