#dev 2020-05-07

2020-05-07 UTC
[chrisaldrich], [Aaron_Klemm], [Kevin_Faaborg] and gRegorLove joined the channel
#
gRegorLove
what is websocket
#
Loqi
It looks like we don't have a page for "websocket" yet. Would you like to create it? (Or just say "websocket is ____", a sentence describing the term)
#
gRegorLove
what is web socket
#
Loqi
It looks like we don't have a page for "web socket" yet. Would you like to create it? (Or just say "web socket is ____", a sentence describing the term)
nickodd, [prtksxna], beko, oodani, swentel, [Bryn_Wolf], loicm, jamietanna, gRegorLove, Mikaela, edrex, dansup, [LewisCowles] and [mapkyca] joined the channel
#
[LewisCowles]
How does a person submit a specification to W3C?
#
sknebel
[LewisCowles]: roughly: W3C specs are done by working groups formed for specific topics. working groups consist of W3C members/their employees + small number of invited outside experts. There is no direct process for "I have a spec and want a W3C stamp"
#
[LewisCowles]
Hmm. Maybe a spec is the wrong idea or wording.
#
[LewisCowles]
I thought spec as it exists outside of any vendor implementation and doesn't tell how to do, just what and why.
plindner[m], jamietanna[m], confraria[m], beko[m], edrex, Salt[m], fredcy_, Rixon, mykiwi, atj[m], bmiller59[m], jceb[m], takev[m], OberstKrueger[m] and ReneM[m] joined the channel
#
sknebel
no, that makes sense. they only produce specs (well, ideally a spec is only finalized once multiple independent implementations have been done, but that's another can of worms)
#
[LewisCowles]
Oh this is a thing with many implementations. It's not "need noteriaty, have cooked up". It exists more or less from pre-internet network technology to now, but I cannot seem to find a spec. I'd much prefer to find someone else to reference and bemoan
#
[LewisCowles]
And possibly learn from, if there is a hidden spec alternative to https://tools.ietf.org/html/rfc7578 I'm unaware of.
loicm, hs0ucy and [jgmac1106] joined the channel
#
[jgmac1106]
how do you mark up an h-card where someone's website has a different name than th author?
#
[jgmac1106]
for now I am just doing title="A Arod Edge Wise" and keeping the p-name for the author name
#
[LewisCowles]
I think this might be a case of "it depends" on the audience you're communicating to
#
[jgmac1106]
in teacher world everybody's blog (like mine) has som cutesy artsy-fartsy name
#
[LewisCowles]
Might I suggest, looking into sr-only CSS classes so that an attribute is not the only place the value is isolated to
#
[LewisCowles]
It provides the following benefits. Text-mode and non-screen viewports, or just devices and situations outside of the one you have now may have cause to interact with that "artsy-fartsy" name in a way that attributes can miss
#
[LewisCowles]
rel="me" is an attribute as it should never be read out to a user
#
[LewisCowles]
but "A Arod Edge Wise" might be something that is a cue for some readers / consumers to de-duplicate or trigger, "oh hey that is {hopefully positive feelings}"
#
[jgmac1106]
this was the blog: http://awordedgewiselindamitchell.blogspot.com/ the author's url or name clearly isn't the name
#
[jgmac1106]
in the end you are right it doesn't matter as nothing consumes my lists of h-cards....though I dream of a day where I add a new h-card the page just gets parsed by a social reader and the blog to the right channel
#
[LewisCowles]
I think familiarity may also have something to do with this. I have a friend Bob who blogs as "Random Tech Thoughts". I generally try to get both names in, using hyphenation, long dashes, commas or separators
#
[LewisCowles]
it gets weird sometimes and I must admit I have no formal way to be sure I'm doing all the time. It's best effort
#
[jgmac1106]
...not that my rss feed and that page match in anyway......
#
[LewisCowles]
I like the attached ID's to headings
#
[LewisCowles]
and that each is clearly a logical context
#
[jgmac1106]
[LewisCowles] when I click on the link I get a .5 maybe faster white screen with a think some code
#
[jgmac1106]
real quick like but jarring flash
#
[LewisCowles]
oh, the non-css HTML
#
[jgmac1106]
yes that's it, no stylesheet
#
[jgmac1106]
had to look again when you said it
#
[LewisCowles]
I think the only way to solve that is to change css naming strategy and serviceworker trying to call network, or to inline all styles on every document
#
[LewisCowles]
I hope it should read without stylesheet.
#
[LewisCowles]
I test every page in a few browsers to try to ensure it will
#
[LewisCowles]
as well as firefox reader mode. I'll look into it
#
[LewisCowles]
I might begin pushing it to a CDN so USA visitors don't see a flash
#
[jgmac1106]
I am on FF Dev not nightly now
#
[LewisCowles]
I think it could be geography and networking rather than the software if it's HTML and not source
#
[LewisCowles]
the offline thing is interesting too
#
[jgmac1106]
yeah I have really bad last of mile connection
#
[LewisCowles]
Although I do fall-back if flaky connection to assuming offline
#
[LewisCowles]
which could explain that too
#
[LewisCowles]
you have < 1s to load any page as it free's up my server to have the service-worker configured that way. It's 30s at nginx I believe
#
[LewisCowles]
I wonder if there is a way to defer my serviceworker without a cookie and without active scripting.
#
[jgmac1106]
yo ever here of HaxCMS? Brian stops by occasionally it is a webcomponents/service worker CMS....he might be able to help....I got nothing
#
[jgmac1106]
ughh think my hands telling me to work a half day today
#
[LewisCowles]
Interesting. I think if it's the Github hosted Biran, I may have looked at it as a source of intrigue. I think I know the path forwards here. Put on a CDN, possibly relax load times to 1s
#
[LewisCowles]
or load, then hot-replace.... IDK, I'll chew on it for a bit
KarlieKloss, swentie, b3u, [Binyamin_Green], dckc, swentel and hs0ucy joined the channel
#
[jgmac1106]
!tell snarfed interesting bug for me on Bridgy, I am getting a message when trying to use flickr imgaes that the file size can't be determined: https://live.staticflickr.com/65535/49857912948_986c454bf4_z.jpg
#
Loqi
Ok, I'll tell them that when I see them next
geoffo, [KevinMarks], [Binyamin_Green], nickodd, jolvera and [snarfed] joined the channel
#
[snarfed]
jgmac1106 sorry for the trouble! twitter has a 5MB size limit on images, so i check the Content-Length HTTP header on input images, but flickr evidently doesn't serve that header
#
Loqi
[snarfed]: [jgmac1106] left you a message 1 hour, 8 minutes ago: interesting bug for me on Bridgy, I am getting a message when trying to use flickr imgaes that the file size can't be determined: https://live.staticflickr.com/65535/49857912948_986c454bf4_z.jpg
#
[snarfed]
i guess i could just go ahead and try the image instead
#
[jgmac1106]
weird snarfed i wonder if it used to, almost all my featured img are hosted on flickr unless I remix mutliple images
#
[jgmac1106]
or more likely this is friends doodle and doesn't have the flize size included...not sure
#
[jgmac1106]
but thx for letting me know
#
[snarfed]
feel free to file an issue!
#
[jgmac1106]
let me try a few more and make sure it isn't a one off edge case....need to be doing some more blog posts anyways
#
[snarfed]
oh it may be an edge case but i can/should definitely still handle it
[LewisCowles] and basil joined the channel
#
Guest7523
hello, does anyone here have any experiance with the webmention.io api?
#
Guest7523
specifically, I have sent a test webmention to my site, which appears in my webmention.io dashboard
[tantek] joined the channel
#
Guest7523
returns
#
Guest7523
{"type":"feed","name":"Webmentions","children":[]}
#
jacky
what's your site URL?
#
jacky
or rather the page you're testing?
#
jacky
what's a reasonable 'minimal' feature set for a hosted webmention service?
#
jacky
right now, I'm thinking 1) a means of checking if a site exposes a webmention endpoint 2) providing a webmention endpoint
#
jacky
keeping history of outgoing would be useful for people but I don't know _how_ important that is for a MVP
#
Guest7523
Hi jacky, all seesm to be working, unsure why. Going to assume user error, typo or something.
#
Zegnat
jacky: outgoing it pretty different from receiving though. As soon as you even mention outgoing there needs to be some sort of way to trigger those outgoing ones, which depending on what you want to support for trigger types can change the MVP a lot
#
jacky
Guest7523: cool!
#
jacky
Zegnat: yeah - that's a good point. I was thinking/hoping the site would do it itself but I could do something like an authorized endpoint that'd, provided some URLs to check for, will send them on behalf of
#
Zegnat
If you are looking at just receiving, I think the MVP is pretty close to what webmention.io is. You add an endpoint on your site for webmention.io and register. Then there is an API that lets you grab all the received webmentions.
#
Zegnat
MVP for sending I guess would be having an API where the user can submit a URL on their own website to, and then you send webmentions for all URLs contained on that page.
#
jacky
hm okay!
#
Zegnat
But for usability you may want to offer some sort of filtering options. E.g. does the user want to send webmentions to themselves?
#
Zegnat
Or which I think Telegraph has as default: only send webmentions for links within an h-entry, ignoring possible links around (thus ignoring menus and other website design elements)
#
Zegnat
I think neither sending nor receiving are really problematic things by themselves. The spec is pretty short. I think the sane user defaults that you want to offer as a service for others is where the real design decisions are at
#
Zegnat
Apropos nothing: I set-up Caddy as a reverse proxy to get local https working on a spoof domain at work today. And wow. I just can’t get over how different this experience was from previous messing with Apache or nginx
#
jacky
that's the part I'm looking more into right now: determining _what_ links to send
#
jacky
I think I might provide a diagnostic page (or a preview) to show for a provided page, what webmentions would be sent
#
jacky
Zegnat: caddy is _amazing_ for that!
#
jacky
hm tbh that diagnostic page might have to double as a light mf2 checker
#
Zegnat
That is also the question: do you care about the mf2 of the page when sending webmentions?
#
jacky
I'll prioritize it since it's the more explicit and cleaner way to do this
#
jacky
but I think falling back to pulling some other stuff and informing the user that it's of "lower quality" b/c of that might be useful
#
jacky
largely to encourage adoption (lol) but also to be candid that using those approaches don't help me help them
Anabella and [asuh] joined the channel; jceb[m] and ReneM[m] left the channel
#
jacky
random: does bridgy (on twitter) respect blocked/muted settings?
#
jacky
ah okay it does!
hs0ucy joined the channel
#
jacky
bridgy++ for having a endpoint to handle registering of users!
#
Loqi
bridgy has 2 karma in this channel over the last year (7 in all channels)
Formaceft and [KevinMarks] joined the channel
#
[KevinMarks]
Mention.tech only looks inside a h-entry too.
#
[KevinMarks]
I think indiewebify.me does all links
#
[KevinMarks]
Mention.tech collects webmentions and passes them through, which is a bit odd in some ways
#
jacky
so I actually want that to be a feature
#
[KevinMarks]
Also it sends all urls (source and target) to Internet archive
#
jacky
like if my site randomly goes down or I have to do some sort of maintenance, I don't want that to be a reason that I lose a webmention (I can't rely on people to retry it)
#
jacky
I think aaronpk does that via werbmention and his site
#
sknebel
if your site is down, the webmention endpoint can't be discovered
hs0ucy joined the channel
#
[KevinMarks]
I was also thinking that you could use it to send them and then you'd have them cached before theybsetbirbup
#
[KevinMarks]
*set it up
#
[KevinMarks]
Right, so send with mention.tech and it keeps them anyway
#
jacky
ha that's actually true
#
[KevinMarks]
But I have not been using this much as I have been relying on my known install instead
#
jacky
see now this makes me wonder if it's wise to do something 'smart' to send webmentions to the last known endpoiint
#
jacky
*endpoint
#
[KevinMarks]
Have a look at the code if you like
#
jacky
for known?
#
Loqi
[kevinmarks] mentiontech: webmention experiment
#
[KevinMarks]
Very appengine centric
gRegorLove joined the channel; nickodd left the channel
#
Zegnat
I am always split on the smart sending. I think some implementations cache an endpoint per domain. But the spec basically says the endpoints exist on a per URL basis
#
jacky
_right_
#
jacky
which, if done with caching, is difficult since I'd have to keep _everything_ lol
swentel, KartikPrabhu, Honey2bee, [Rose], [tw2113], leg, [Aaron_Klemm] and [snarfed] joined the channel
#
[snarfed]
yup. bridgy's per-domain caching is useful for scaling. it notably distinguishes http vs https and home page vs all other URLs though
#
jacky
hmm how do you determine / define a home-card? like the page that holds one's representative h-card
#
jacky
*h-card?
#
jacky
(that was a question lol)
#
jacky
what is homepage
#
Loqi
Your homepage represents you on the web, typically at the top of your domain, with your name and an iconic representation, often marked up with h-card, and fairly commonly one or more streams of recent, topical, or most relevant posts marked up with h-entry https://indieweb.org/homepage
#
Loqi
[snarfed] #4 do we need an h-feed authorship algorithm?
#
jacky
oh I mean like how does bridgy rationalize a 'home page'?
#
jacky
like just by its root domain?
#
jacky
(i.e: https;//cookies-are-awesome.com/)
#
[snarfed]
oh yeah home page is just / path
#
[snarfed]
afaik bridgy doesn't currently use either representative h-card or authorship
#
jacky
gotcha!
[KevinMarks], [schmarty], KartikPrabhu, [fluffy] and gxt_ joined the channel