#dev 2020-03-09

2020-03-09 UTC
[KevinMarks] joined the channel
plut4rch joined the channel
#
vilhalmer
I was ready to be upset if this loaded blank without js
djmoch_ joined the channel
#
[Michael_Beckwit
except it's all JS driven, not js sprinkled on as garnish. the js renders everything for you now
djmoch and gRegorLove joined the channel
#
[tantek]
what is jamstack
#
Loqi
JAMstack is a web development architecture based on client-side JavaScript, reusable APIs, and prebuilt Markup https://indieweb.org/JAMstack
#
Loqi
ok, I added "https://paramaggarwal.substack.com/p/how-websites-evolved-back-to-static-html-css-js-files-57ce549f81db" to the "See Also" section of /JAMstack https://indieweb.org/wiki/index.php?diff=68907&oldid=42432
pablinos, [ColinMorris], superjen96, beko, [tantek], [LewisCowles], plut4rch, nickodd, paulbunkham, jjuran, cweiske, [KevinMarks] and Kopsi joined the channel
#
Kopsi
Hi!
vika_nezrimaya and [g33kcentric] joined the channel
#
Kopsi
I am researching IndieWeb for my masters thesis and was curious, whether there have been any recent privacy/security surveys of it. Google hasn't provided any results regarding this, so i was curious, whether there was something unpublished or published in a way that couldn't be found by Google.
#
vika_nezrimaya
Privacy/security surveys? what do you mean?
#
vika_nezrimaya
Kopsi: Also I think you're better off in #indieweb with this kind of question, this chat is more hardcore technical stuff
nickodd joined the channel
#
Kopsi
Okay, thanks for the input
#
Kopsi
With security/privacy survey I meant a structural analysis of the components regarding security and privacy implications and possible weaknesses. I did a small survey of a specific IndieAuth implementation recently and (in theory) have found two minor vulnerabilities based on OAuth vulnerabilities. So thats the sort of stuff I was looking for.
[jgmac1106], nele1, [KevinMarks] and hs0ucy joined the channel
#
GWG
I am finding a lot of multi author markup in converting JSON-LD to Microformats 2
#
[KevinMarks]
what's the source of the json-ld?
hs0ucy and petermolnar_ joined the channel
#
vika_nezrimaya
I'm interested in one android thing and while I doubt that someone here can answer, I'll still drop this here: if Android places background restrictions on apps, then how the heck Google Play Services evades them and still manages to, e.g. deliver push notifications to apps using FCM? Is this restriction simply waived for apps in /system/priv-app? Documentation obviously says nothing about it so people can't replicate what GMS does to
#
vika_nezrimaya
stay alive
#
vika_nezrimaya
I want to receive notifications for Kittybox but a foreground service notification is simply too ugly :c while it can be hidden, it's still a bit ugly
#
vika_nezrimaya
and no google play services allowed, this is a FOSS app :c
pablinos joined the channel
#
[KevinMarks]
I think Google Play gets background approved by default
#
vika_nezrimaya
the question is how could I inject an app in the ROM the same way Google does
#
vika_nezrimaya
aha there is /system/etc/sysconfig that seems to whitelist some apps, including google ones
[grantcodes] joined the channel
#
[grantcodes]
There are a few apps that just show a message when they are battery restricted and tell you to change their settings
TheCondor, [LewisCowles], aaronpk, jkphl, [jeremycherfas], [yo], [schmarty], [Michael_Beckwit, geoffo, [email096], plut4rch, [snarfed], nickodd, gRegorLove, [tantek], dougbeal|iOS, uniquerockrz, pablinos, [jgmac1106], KartikPrabhu, leg and lalilulelo joined the channel
#
[jgmac1106]
There was a change to archive.org around embedding media and I can't find an answer: https://archive.org/post/1106251/replacing-details-and-downloads-no-longer-works-for-embedding-video-and-audio
#
[jgmac1106]
basically you used to be able to embed audio and video using the media src url and HTML...now it seems some are arbitrarily not allowed...but no documentation on the change anywhere
#
[tantek]
I wonder if it's worth adding a rel=canonical validator to the h-entry validator in IndieWebifyme
#
[tantek]
per [jgmac1106] tweet and gRegorLove correction (moving here from #indieweb channel)
#
[tantek]
thoughts?
#
[tantek]
My thinking is that it would be nice to point someone to a validator explaining why their rel=canonical is malformed rather than give them one-off uncited advice
#
gRegorLove
interesting. Would it check the content type in this example, e.g. "looks like you linked to an RSS feed"?
#
gRegorLove
suggesting adding rel-canonical if none found I think would be a good idea, though
#
[tantek]
MVP would be validating the href (that it loads 2xx)
#
[jgmac1106]
to note John was referring to another person's website and not mine, the issue arose as I get annotations mapped to the wrong post often
#
[tantek]
hmm I wouldn't recommend adding it unless there was a specific reason to
#
[tantek]
typical case is someone checking a post on their own website which doesn't need rel=canonical
#
[tantek]
if it's going to suggest adding some markup, it should provide a specific impactful reason to do so
#
gRegorLove
fair point
#
[tantek]
we should break people of the habit of adding voodoo markup like Dublin Core
#
gRegorLove
maybe if there's a syndication link found?
#
[tantek]
nope still don't need canonical for that
#
[tantek]
syndication just means the *other* thing is a copy
#
[jgmac1106]
ahh it is happening right now: https://quickthoughts.jgregorymcverry.com/ that top post wasn't annotated
#
[tantek]
if there is a valid rel-canonical link, we could suggest a one-click link to check that validate that post's h-entry
#
gRegorLove
Right, but it indicates there's two copies of this out there, so you should probably mark yours (original) as canonical
#
gRegorLove
*at least two
#
[jgmac1106]
this one was...hypothes.is supposed to use rel="canonical" to match...we think someone annotated the post on the feed page and not url of the link (someone being me)
#
[tantek]
self rel canonical has no impact
#
[tantek]
it's like vouching for yourself
#
[tantek]
"I'm really a nice guy, really"
#
gRegorLove
Hm, I'm missing something
#
[tantek]
note, the h-entry validator is only validating *one* post
#
[tantek]
that's a completely different project
#
[tantek]
so it can't assert anything about those other posts
#
[tantek]
not like a whole network of posts linked to from that post
#
gRegorLove
Ok, backing up a bit... if there are 2 URLs with the same content, I thought the ideal setup would be to link rel-canonical on both of them. Is that not correct?
#
gRegorLove
Obviously with a lot of POSSE that's not an option, so I was thinking next best thing is rel-canonical on the URL you control
#
gRegorLove
"rel-canonical on both of them"... meaning pointing to the single, canonical URL you control
#
[jgmac1106]
on the rel canonical link I add an also on "u-syndication" on the syndicated copy I add the also on rel ="canonical"
#
[tantek]
AFAIK it is is a waste of markup to make something rel-canonical link to itself
#
[tantek]
so "rel-canonical on both of them" makes no sense
#
[tantek]
not sure how else to say that
#
gRegorLove
Example: my photo post here https://gregorlove.com/2020/02/yo-dawg/ is syndicated to Instagram. If a search engine indexes both, I want them to have mine be canonical. I can't specify rel-canonical on IG, though.
#
Loqi
[gRegor Morrill] Yo dawg, we heard you like cards so we put a card inside your card. The small card is about 3 inches and was inside its own tiny envelope. Many thanks to Isha for these. <3 #abracaDAYUM https://gregorlove.com/site/assets/files/5759/img_20200215_135941.1000x0-is.jpg
#
[tantek]
I understand what you want. What I'm saying is rel=canonical linking to yourself makes zero impact AFAIK towards what you want
#
[tantek]
because again, self-vouching is ignorable
#
[tantek]
for the same reason as the "Believe me... " "Trust me..." type statements are empty
#
gRegorLove
Google indicates they choose the canonical page for search results, and crawl them more freqauently, in part based on rel-canonical
#
Zegnat
Self refering rel-canonicals are recommended if the same resource is available with multiple different query parameters. IIRC from my SEO sessions. So if you have something like example.com/article?source=facebook and example.com/article?source=twitter and want search engines to index both as just example.com/article it is recommended to have that as a rel canonical
#
[tantek]
then it's no longer "literally" self-referring. it's referring from a version with rando query params to the non-query params version
#
jacky
wait so to catch up
#
jacky
using rel=canonical on a site to point back to one's original post to encourage proper 'ownership' of posts?
#
[tantek]
rel-canonical only ever makes sense when from one URL to a different URL
#
jacky
that's neat
#
Zegnat
Yep, but that is the only context in which I have heard “self referring” used
#
jacky
I'm guilty for self-canonicalizing content :( lol
#
[tantek]
otherwise it's a waste of markup, not unlike DC
#
Zegnat
I have definitely slaped some rel="canonical" on <a> elements that I had pointing to the resource itself, because “why not”. But do not expect them to do much.
#
[tantek]
"why not" because noise
#
gRegorLove
I would have to write *more* code to only include the rel-canonical on non-self-referring pages :)
#
[tantek]
because any addition to page weight actually *harms* your info density and search engine evaluation
#
[tantek]
especially invisible additions like that
#
Zegnat
“ and search engine evaluation” not according to the people we pay to check that.
#
[tantek]
right, same people that encourage all the rando markup you don't need
vasilakisfil joined the channel
#
[tantek]
of course they provide more markup, their job is to convince you they are providing something of value, not to actually provide value
#
Zegnat
Except that when we added Schema, we went up on result pages, so that mark up was apparently needed to compete with sites that also have it :P
#
vasilakisfil
hey folks what do you think about jitsi ? I mean from indieweb/open protocols perspective
#
[tantek]
it's like paying for the underswash
#
sknebel
vasilakisfil: it's not bad
#
[tantek]
Zegnat, for example if they provide DublinCore you know they are "selling you something"
#
[tantek]
snakeoil as it were
nickodd left the channel
#
Zegnat
Oh, sure, there has never been a mention of DC. And I do not think I have ever seen anyone say they parse DC
#
sknebel
We used it for a while for indieweb community things, but decided zoom served our use cases better
#
[tantek]
SnakE Oil
#
Zegnat
We also do the ownership thing through rel-canonical, kinda. When one paper shares an article from the other, it gets displayed in their design, but with a rel-canonical to the paper the article first appeared in.
#
Zegnat
But I have not heard whether that has been boosting us in anyway (yet).
[KevinMarks] joined the channel
#
[KevinMarks]
If you rel-canonical to the bare url it works if people link to you with query params or fragments
#
[KevinMarks]
Which sites often do
KartikPrabhu joined the channel
#
[KevinMarks]
Twitter will add an ?amp others add utm_cruft
#
Zegnat
Yep, that is why we do it at work
#
gRegorLove
Yeah, that's another important use-case
#
Zegnat
And as gRegorLove said, it is often more code to first detect the cruft and then render the rel-canonical, so you just end up always having it there
#
[KevinMarks]
Especially if your site is static
#
@ChrisAldrich
↩️ I maintain that if the source of the article and the target both supported Webmention, then when a piece used an image with a link, then the original source (any website, or Flickr [more...] @brainpicker @swissmiss https://boffosocko.com/2020/03/09/idea-a-script-to-find-flickr-photos-being-used-online-matt-maldre/
(twitter.com/_/status/1237101274094374912)
[chrisaldrich] and pablinos joined the channel
KartikPrabhu and Mikaela joined the channel
#
[tantek]
Zegnat, gRegorLove you're right that it is more work to detect, I'm only detecting one non-canonical version and doing something special. If you load any of my pages with http:, you get a rel-canonical to the full https: URL
#
[tantek]
otherwise if you load my posts with https, then I use a path relative rel-canonical URL
#
[tantek]
I suppose I could conditionalize the whole thing on presence of http: or presence of query param(s)
KartikPrabhu joined the channel
#
gRegorLove
I was just noticing the relative rel-canonical on one of your posts. Interesting
#
Zegnat
Yeah, if you think that code complexity is worth it to get away from “any addition to page weight”.
#
[tantek]
yes I think minimizing markup is useful (there's lots more that I could minimize), for slow clients, for teaching (less noise on View Source), and lastly for info density for search indexing
#
[tantek]
pretty sure it's like a one line test
#
gRegorLove
I don't follow though, isn't the rel-canonical here self-referring once you use the base href? https://tantek.com/2020/068/t4/teresa-keynote-more-civil-social-web
#
Loqi
[Tantek Çelik] Freshly posted: Toward a More Civil and Social Web @optoutools founder Teresa’s keynote @IndieWebCamp Berlin last November 12:10s video: https://archive.org/details/iwcberlin2-keynote-teresaingram #AI #AIEthics #NLP #machineLearning #socialWeb #s...
#
[tantek]
right that's what I'm fixing now 🙂
#
@mattmaldre
↩️ As @Flickr is trying to find ways to get people to subscribe to their service, including a feature like webmentions in their service would be wonderful. As a photographer, I would LOVE to know when sites use my photos. If Flickr could facilitate that, AWESOME.
(twitter.com/_/status/1237112636552200193)
#
[tantek]
changed an else to
#
[tantek]
else if ($_SERVER['QUERY_STRING']!='')
KartikPrabhu joined the channel
#
[tantek]
e.g. try that same URL you cited above gRegorLove (which should no longer have any rel-canonical), then try this version: https://tantek.com/2020/068/t4/teresa-keynote-more-civil-social-web?skin=trn
#
Loqi
[Tantek Çelik] Freshly posted: Toward a More Civil and Social Web @optoutools founder Teresa’s keynote @IndieWebCamp Berlin last November 12:10s video: https://archive.org/details/iwcberlin2-keynote-teresaingram #AI #AIEthics #NLP #machineLearning #socialWeb #s...
#
gRegorLove
makes sense
#
[tantek]
I'm deliberately including the relative path rel-canonical to be a test case for rel-canonical parsers / scrapers
#
[tantek]
in case any of them were/are assuming absolute URLs (which would be against the HTML spec)
#
[tantek]
also it's slight shorter code 🙂
#
gRegorLove
true, since you're already using base href
#
[tantek]
yeah that makes it an extra fun test
[yo] joined the channel
#
[yo]
something to consider for those of us providing API services (or planning to): https://www.rdegges.com/2020/the-only-type-of-api-services-ill-use/
[Michael_Beckwit joined the channel
#
wink
I don't see this being very relevant
#
GWG
I'm wondering about consuming formats I don't publish being hypocritical.
[LewisCowles] joined the channel
#
sknebel
don't think it is
#
sknebel
just because other people think they are the right choice for them to publish (which presumably is driving you writing consuming code) doesn't mean you have to think that
#
GWG
I'm a purist in the sense I convert it all to microformats
#
wink
it would be hypocritical if you'd preach to the masses and don't do it.
#
GWG
Which it?
#
[tantek]
nah, see Postel's law
#
GWG
What is Postel's law?
#
[tantek]
!wp Postel's law
#
[tantek]
darn I thought that might work
#
GWG
I'm surprised Loqi didn't answer
#
[tantek]
enwp.org/Postels_law
#
[LewisCowles]
I agree with the comment that Postels law can be dangerous, but really like the CSS take on compliance. Just try to make as many lines as you can parse work
#
[tantek]
CSS has made a good balance with forward/backward compatibility in great part due to open and fairly thorough test suites (which helps avoid (most) of those "bug for bug" compat problems. at least minimize them. a few still pop-up time to time.)
[Joe_Crawford] joined the channel
#
[tantek]
I skipped it because it was compared to a bad reinvention of the STYLE attribute
#
[KevinMarks]
I may be missing some nuance but he seems to have gone the long way round to inline style attributes, but with worse syntax.
#
[KevinMarks]
He's citing a lot of css people I do pay attention to. I'm feeling like a Poe's law thing like Jenn Schiffer does (he invents a <Box> tag ffs. But I think he might be serious.
#
[tantek]
He cites some good inspiration, and then goes off into the weeds and reinvents effectively inline Style attribute styling because he doesn't like to name things, thus missing all the points about reusable components etc. in order to re-use them, you have to name them.
#
[tantek]
so it's pretty ignorable IMO
#
@StuRobson
designer: can we update the card component to use the largest padding? developer: the card component? Do you mean – rounded-lg my-2 p-4 justify-center?
(twitter.com/_/status/1235472931528028161)
#
[tantek]
the bottom of the post links to Diana who recently retweeted this which illustrates the problem with pretending you don't have to name things: https://twitter.com/StuRobson/status/1235472931528028161
#
[KevinMarks]
Literal architecture astronautics “Theme UI is the next step up this ladder of abstraction”
hs0ucy joined the channel
#
[tantek]
encapsulation and components are still a challenge in the web platform. I think if those were solved it would address a lot of the things that frameworks are working far too hard to do
KartikPrabhu joined the channel