#dev 2020-06-28

2020-06-28 UTC
#
[fluffy]
of course in my feed (and posts) I have it visually call out when a post is granted via privacy rules
lahacker joined the channel
#
[fluffy]
it puts a lock emoji in the title, the feed item gets a “Note: this is a private post, please use discretion in sharing it,” and it also gets an extra CSS class that turns the title purple when viewed directly on my site (which obviously doesn’t apply to the social reading experience)
#
[fluffy]
Hopefully that came through in my brief demo 🙂
#
jacky
I did see that!
#
jacky
affordances++
#
Loqi
affordances has 1 karma over the last year
#
[fluffy]
I mean I never showed what happens in the atom feed or anything but that’s all just up to my templates
jonnybarnes and [chrisaldrich] joined the channel
#
[fluffy]
yeah I was using unlisted entries for a little while, until I had auth working
#
aaronpk
i want this to be so much easier
#
[fluffy]
I still wasn’t comfortable with the security tradeoffs on that, and thre was still the sideband notification issue
#
[snarfed]
oh sure, it's a spectrum, some stuff has to be actually auth-based private
#
[snarfed]
but in practice that immediately excludes the vast majority of people, who will never log into whatever auth system we're building
#
[snarfed]
at least for my audience
#
[fluffy]
yeah, this is why I built Authl specifically so I could support as many silo auth mechanisms as possible
#
[fluffy]
but that only solves the “how do I see this” aspect, not the “what is there to see” aspect
#
aaronpk
i don't think we emphasized this enough in the session but i feel like the "but they will never log in" problem is just a small difference from "but they will never visit my site" problem
#
aaronpk
where the solution is a reader, so the content comes to them
#
[fluffy]
yeah that’s fair
#
aaronpk
which we know is a useful pattern
#
[snarfed]
eh, maybe. unlisted is definitely an impermanent hack until if/when everyone does this...but until then, my audience is no more likely to use an indie reader than they are to indie log in
#
aaronpk
i guess i'm more thinking like of the people who do use readers, let's at least make sure those readers can support private content
#
[fluffy]
so, as a data point: of the people who have authenticated to my site since whenever the auth table drops off, 18 used an emailed magic link, 3 used twitter, 6 used Mastodon, and 5 used IndieAuth
#
aaronpk
makes sense, i think email notification + email login is going to stick around as a reliable solution for a long long time
#
[fluffy]
a bunch of the emailed links were to people not in my auth list; pretty much all of them seem to have done that in order to get at my GRS photo collection, which I just require people to login to see.
#
[snarfed]
aha yeah email magic link works
#
aaronpk
oh interesting
#
[fluffy]
there were also a handful of twitter auths on that one
#
[fluffy]
actually yeah that makes for some more interesting metrics: of people viewing that entry recently, 13 are email, 1 is twitter, and 1 is Mastodon, and only the mastodon user is a friend of mine (they are getting the same procedure in a few weeks)
#
[fluffy]
there’s actually only one auth failure in the past month, and that’s [chrisaldrich] who never logged in after I granted him access to the entry that I linked here 🙂
#
[fluffy]
well, s/logged in/reloaded/
#
[chrisaldrich]
fluffy, I made a note to come back to that post at a later date, but what was missing was a notification that I had been granted access. Or if there was one, I missed it.
#
[fluffy]
Most of the folks who see my private entries are folks who follow me on Mastodon and see the stub post.
#
[fluffy]
there’s no notification aside from me telling people they’re authorized. I thought I messaged you!
#
[fluffy]
I thought I’d @-notified you here, I mean, but I may have forgotten.
#
[fluffy]
For privacy granting stuff I definitely prefer it to be a conversation rather than a protocol.
#
[fluffy]
anyway on that particular entry about half of the people who looked at it were folks I’d specifically given a sideband notification to, and the other half saw it through whatever mechanism they use to follow me normally
test, paulopinto and jonnybarnes joined the channel
#
[schmarty]
gRegorLove: any interest in adding a crawler and feed follower to indiebookclub? These days I often post reads via Shortcuts but I have a bunch I posted with ibc. Would be neat to combine!
#
[schmarty]
(I have some interest in working on it but would need to know some details about how the site is hosted)
#
gRegorLove
Hm, not sure I follow, but happy to discuss. Ryan and others kinda swayed me towards possibly setting up some discovery stuff... maybe tomorrow.
#
Loqi
[gRegorLove] #9 Consider aggregating read posts via webmention
#
gRegorLove
[schmarty] want to talk in the hallway tomorrow? My brain is pretty tired at the moment (plus I know it's getting late there)
#
[schmarty]
Yes please! I need to review the reading session notes anyway!
[jacky] joined the channel
#
[jacky]
👁👁
#
[jacky]
comes into the hallway
#
jacky
I didn't see the word 'tomorrow'
#
KartikPrabhu
jacky: aaronpk and I are in the hallway
#
jacky
lol gonna take a quick nap tbh
#
jacky
gotta make dinner
#
GWG
I took a 20 minute nap
#
GWG
Are people still on?
#
aaronpk
hallway party
#
gRegorLove
I took a nap towards end of 2nd session. was so tired today
#
GWG
gRegorLove: We all need sleep
jonnybarnes, gxt_, nickodd and treora joined the channel
#
Loqi
xi-editor retrospective
jonnybarnes joined the channel
#
@edent
What's the semantic way to say in HTML "this work was cited in that academic paper"? I know how I cite something in SchemaOrg (see screenshot) - but I want to do the reverse. Like a pingback or webmention. Is this possible / sensible / useful?
(twitter.com/_/status/1277113950598029317)
geoffo, jonnybarnes and [Chaitanya] joined the channel
#
[Chaitanya]
Is it possible to add CSS styling to the title attribute in a similar way to this a[data]:hover:after ?
jonnybarnes joined the channel
#
[Chaitanya]
If possible a very simple multi-line styled hovercard (without images) can be made with the anchor element. Pretty basic I know but looks good in a way 😅
jonnybarnes, moppy and [fluffy] joined the channel
#
[fluffy]
I don’t think there’s any way to style the HTML title popup thing but CSS’s attr() function and content: property can do some pretty neat stuff 🙂
#
[fluffy]
My day job’s frontend has a bunch of clever things with that to make responsive table layouts
gRegorLove and jonnybarnes joined the channel
#
[Chaitanya]
fluffy++ Thank you! I am going to try that out now.
#
Loqi
fluffy has 21 karma in this channel over the last year (72 in all channels)
geoffo, jonnybarnes and jamietanna joined the channel
#
jamietanna
Are there any services that, after sending webmentions to i.e. Bridgy, will then send a Micropub update with the syndicated post's URL? I've got this set up on my own post-deploy infrastructure, but it'd be good to find out if there's anything else there that could do it
jonnybarnes joined the channel
#
ben_thatmustbeme
that is a really interesting idea
[jgmac1106] joined the channel
#
jamietanna
I've done it on my side because I've got a static site, and I need to publish the site first to be able to send the WM to Bridgy to then get the final syndicated URL, which isn't as necessary for dynamic sites
#
@jgmac1106
↩️ I use h-cite on my vita mark up with microformats https://jgregorymcverry.com/resume wou be neat if webmentions would appear between 2 publications. X was cited in Y and then on my author page I get a webmention noting someone cited me (https://quickthoughts.jgregorymcverry.com/s/20VNFs)
(twitter.com/_/status/1277198472882552834)
geoffo joined the channel
#
sivy
[tantek] do you have any notesfrom your thoughts or discussions you could share?
#
sknebel
[fluffy] aaronpk and others: I'm probably working on other things today, and was too tired to really participate in the discussion yesterday, but if someone wants to talk AutoAuth specifically I'm available (either here or quick zoom call)
rebelle and [schmarty] joined the channel
#
ben_thatmustbeme
hmm, i'm feeling the need to throw away everything and start from scratch with something totally new
#
ben_thatmustbeme
seems a little foolish.... or is it
#
jgmac1106
ben_thatmustbe kinda what I did when I walked away from WP…I keep Known and micro.blog going to provide the stuff I don’t know how to do yet
#
aaronpk
I did that with my last site rebuild and it went great! Still pretty happy with how it went and I don't feel like I need to throw out my current site 4 years later!
#
jgmac1106
on my main domain I am been waffiling of continuing to just write the pages, try Kirby again, or follow the Eleventy bandwagon
#
jgmac1106
^^ might keep with slow slog for what aaronpk notes but hit a point where I really have to learn now
#
jgmac1106
but like I publised a new article…it is a new month I need to make a card for that article add it to the homepage and then make an article page for this month and then copy the article there…..this kinda stuff annoying me and making me not write as much
#
ben_thatmustbeme
i'm thkinking i'm going to try laravel again, but this time try to make everything as installable modules
#
aaronpk
The one downside with using Laravel is they update it so often
#
aaronpk
so I'm on version 5.2 which is way out of date now
#
aaronpk
but so many internal API things changed that I'm worried about updating it
#
ben_thatmustbeme
at least their update docs are pretty clean
#
ben_thatmustbeme
but yeah, it is annoying maintinance
#
ben_thatmustbeme
especially when half the things they update aren't really fixes or anything, they are just changing things around or switching to newer libraries
#
ben_thatmustbeme
but with 5.5 they are supposed to be doing LTS versions now
#
ben_thatmustbeme
its either that or i go back to my super crazy hack of opencart, which is very open, but i have to do everything, there are no libraries. I mean, its way out of date, but its not like the changes they made actually are for security, the things that are left from opencart is very very little
#
ben_thatmustbeme
i'd love to work with someone on something though too
#
ben_thatmustbeme
need to find a PHP person in here who wants to get a cook for others system going
jonnybarnes joined the channel
#
ben_thatmustbeme
i guess it doesn't have to be PHP, just my strong suit
#
aaronpk
Yes I'm kinda sad I started just early enough to miss the LTS versions
#
ben_thatmustbeme
why not just make a full clone and start updating locally as a test
#
ben_thatmustbeme
the docs for upgrade pretty much hand-hold you through them
#
aaronpk
cause I have a long list and this results in no visible benefit
#
ben_thatmustbeme
as i rememnber 5.2 - > 5.3 -> 5.4 was super easy
#
ben_thatmustbeme
it was the 5.4 -> 5.5 that was more involved
KartikPrabhu joined the channel
#
aaronpk
and now they're on 7
nickodd joined the channel
#
aaronpk
they do a reasonable job of documenting things to upgrade but it still just sounds like a lot of not fun work
#
aaronpk
still overall I don't regret using laravel for this tho
#
ben_thatmustbeme
well, they changed their numbering
#
aaronpk
5.2 -> 5.3 -> 5.4 -> 5.5 -> 5.6 -> 5.7 -> 5.8 - > 6.0 -> 7.0 sounds like a lot of upgrades, I wonder if I'm better off starting with a fresh 7.0 install and porting my code into it instead
swentel joined the channel
#
ben_thatmustbeme
possibly, but i think the upgrading will actually be easier
#
aaronpk
yea a lot of the changes won't affect me because I use relatively little of the framework
wombelix and jonnybarnes joined the channel
#
ben_thatmustbeme
i feel like most of the work will be in verifying that everything is still working at the end
#
aaronpk
yeah and I don't have tests :(
jonnybarnes joined the channel
#
[schmarty]
Just felt this sting for a project that was on Rails 4 and getting security alerts on GitHub with no patches available
#
[schmarty]
Ended up migrating to rails 5.2 (barely adequate since it only gets security updates). Finished in a few hours but had to touch more things than I expected
#
[schmarty]
Also feeling this a lot at work where we are trying to get off php 7.1 for some things 😭
[LewisCowles] joined the channel
#
[schmarty]
All that to say I am feeling more than ever that no code is better than no code 😂
#
aaronpk
sometimes i dream about deleting all the code i've ever written
#
swentel
I often do that with issues
#
swentel
just delete them all, must say that helps :)
#
aaronpk
also my email inbox
[manton] and [tantek] joined the channel
#
[tantek]
sivy, context? All notes from yesterday’s sessions are linked from /2020/West/Schedule
#
[tantek]
Lol Aaronpk
#
jonnybarnes
A client makes a get request for the latest media uploaded to the media endpoint, there isnt any, then return empty json with 404 status?
#
jonnybarnes
is what I’ve just done for my endpoint
#
aaronpk
404 doesn't sound right
#
aaronpk
that would suggest the media endpoint is gone
#
GWG
aaronpk: We never got to some of the Media Endpoint Micropub issues proposed.
#
jonnybarnes
hmm, so before Quill was showing `null` for images
#
aaronpk
true! impromptu micropub issue session on day 2? it's like a joint project
#
GWG
aaronpk: Happy to do it.
#
GWG
I have more to do on the page, which might be one of my projects for today
#
aaronpk
i'm almmost done with my first project so i'll at least have something small to demo :)
#
GWG
I'm also adding new labels to the Micropub-extensions repo to track
#
GWG
I put in Stable(but discussion required on parameters), Needs-Clients, Needs-Endpoints
#
GWG
Figured it was a good way to encourage people
[chrisaldrich] joined the channel
#
[tantek]
Woke up with a bunch of things from yesterday and recent UX discussions remapped together
jonnybarnes joined the channel
#
jonnybarnes
so if I shouldn’t return 404, what should I return? and if there are no media uploaded, should the response still have a url key, but woth on value?
#
[tantek]
I’m convinced now that this is a spectrum:
#
[tantek]
mute all posts ... mute some types of posts ... see “interesting” posts ... see all posts ... notify for “interesting” posts ... notify some types of posts ... notify for all posts
#
[tantek]
For each person you are following / subscribed to
[KevinMarks] joined the channel
#
[tantek]
More finer grained variants are possible of course
#
[tantek]
Yet this is a good start on a “knob” you could dial up or down for particular people depending on how often they post, what kinds of things they post (that you’re interested in or not etc)
#
@ChrisAldrich
↩️ I've noted before how Altmetric does what some would call backfeed, though I'm not sure what or how their mechanism works other than some heavy search and extreme processing from social media platforms. @jgmac1106 [more...] https://boffosocko.com/2020/06/28/outline-for-webmentions-in-conjunction-with-academic-citations/
(twitter.com/_/status/1277273587594850305)
#
aaronpk
let me check what mine does
#
[tantek]
For “types of posts” consider that in a generic sense, not just strict “post types”, eg posts with (or without!) particular hashtag(s) etc
#
[tantek]
So that’s the “what do I want to see from this person, and how urgently?” axis
#
aaronpk
jonnybarnes: looks like mine returns url: null with a 200 response.
#
jonnybarnes
sounds reasonable
#
[tantek]
There’s another axis which is the complement, “what do I want to *allow* this person to see, and how close to real time?”
#
[tantek]
Applies to both the notification relate and the private posts related conversations we had yesterday
#
jgmac1106
staring here to make my bookshelf: https://codepen.io/smashingmag/pen/xxGKLZO wanna do somehting like this without the js: https://pie-rhinestone-belly.glitch.me/
#
jgmac1106
not sure why the animation stopped workin gin that anyway
#
jgmac1106
when I taught 7th graders in our New Haven Camp I would send parents 5 vocab words a week through SMS to play with their kids
#
jgmac1106
kinda makes me want a subscribe by sms feature…my sureest way to reach people in lofi low data homes
#
[tantek]
^^^ #indieweb-meta ? Good tip for being age-inclusive?
#
[tantek]
Ah, sms notifications are a separate thing, and prior art at Twitter
#
aaronpk
needs to remember to check his dev site instead of production site when making changes in dev
#
[tantek]
Specifically “subscribe by sms” - pretty sure you could *only* do that when Twitter started
#
aaronpk
oh no what did I do. I broke everything
#
aaronpk
oh I was typing in the wrong domain
#
aaronpk
is on a roll
#
aaronpk
needs more coffee
#
jgmac1106
I think petermolnar will know of some zapier sms integrations
#
jgmac1106
struggling trying to get grid to work in the card flip gonna try sticking with flexbox but using a vertical direction
#
[tantek]
Aaronpk, this is a good reason to go for a run to get coffee before camp day 2 officially starts!
#
aaronpk
isn't risking venturing outside unless he has to
#
petermolnar
I was summoned
#
Loqi
[VeronaLabs] Description By WP SMS you can add the ability of SMS sending to your WordPress product. So you can send SMS to your newsletter subscribers or your users and get their attentions to your site and products. Using WP SMS you can enjoy many features, Yo...
#
sivy
[tantek]: i was talking about notes on webmention storage
#
sivy
sorry, old "thread"
#
petermolnar
ah, sms. Yeah. Summary: gammu with linux works great, but setting it up as a webhook is a decent amount of work; with that, you have your own SMS server. Old android can not be used as SMS gateways. (Unless OLD eg. 2.3, and you still have working apks for them, let's not go there). Anything else will cost you.
#
petermolnar
so if someone wants to POSSE to DM, telegram or discord is a better way
#
[tantek]
sivy, indeed worth collecting those notes.
#
[tantek]
Did you explore /2019/Schedule to look for notes from last Summit?
#
[tantek]
Also I want to broaden the scope from “webmention storage” to “storage of other people’s stuff”, eg including reply contexts, icons etc
#
sivy
no, I did read one of the brainstorming docs
#
[tantek]
I think storage of all those “external” things deserves very different (partitioned) treatment HSM storage of your own content
#
[tantek]
than* storage (lol autocorrect)
#
sivy
HSM?
#
[tantek]
(HSM is Haight Street Market, my local grocer)
#
[tantek]
autocorrect privacy leak!
#
sivy
but yes, figuring out how to partition that content is my struggle right now. I’m still using a jekyll-style “yaml+markdown” file system
#
[tantek]
Pretty mild but I can imagine that being A LOT worse!
#
sivy
ah, s/HSM/than/ :D
#
[tantek]
sivy, one thing for sure, gotta store external content as “raw” as possible, and only use stuff like Yaml or JSON representations as caches
#
sivy
I’m considering Hugo.Jekyll style data files, JSON or YAML stored in a “data” directory”
#
sivy
yes, I’m definitely storing the source content
#
sivy
along with meta
#
[tantek]
So Eg Webmentions, gotta store the source HTML (perhaps up to a size limit?)
#
[tantek]
Then don’t mix caches with storage
#
[tantek]
One is purgeable / reconstructable. The other is not
#
[tantek]
So “store data files JSON or YAML” makes no sense
#
[tantek]
Call them caches because that’s what they are and treat them as such, not storage
#
[tantek]
Only exception to that is perhaps if you do you’re own manual editing of others’s reply contexts or webmentions, then it becomes mixed multi source content
#
[tantek]
your* own manual
#
sivy
eeeeenh, not sure I agree, but :shrug:
#
[tantek]
Gah autocorrect ok giving up this iOS typing for a bit 🙄
#
[tantek]
caches != storage; you’ll regret it if you mix that up
#
sivy
in my case, the FS is the system-of-record, my DB is just for speed of rendering/searching
#
sivy
ie, the DB should be able to be rebuilt by reading/indexing the filesystem
#
jgmac1106
grrrr…I can’t get the flex or gird layout I want: https://codepen.io/smashingmag/pen/xxGKLZO
#
jgmac1106
maybe I lose the title and the author and put it in alt text
#
sivy
[tantek]: yeah, I’m liking my fs storage less and less these days
[Beto] and [grantcodes] joined the channel
#
[tantek]
sivy, just wait til you have to backup or migrate, then every other storage system tends to break
#
[tantek]
or your db sw is updated and breaks compat
#
[tantek]
fs storage is good for decades of resilience
jamietanna joined the channel
#
jacky
sknebel: you around re: chatting on AutoAuth?
[hibs] joined the channel
#
sknebel
jacky: gimme ten mins
#
jacky
make it fifteen :)
gRegorLove joined the channel
#
ben_thatmustbeme
looks around for schmarty
#
sknebel
jacky: am back, ready when you are
KartikPrabhu and geoffo joined the channel
#
sknebel
guess we can get a breakout room from the general hangout?
#
jacky
yeah; that'd work
#
jacky
sorry was making something to eat (got up late today)
#
jacky
but tbh it's just one question that I didn't see on the Wiki
#
jacky
did https://www.rfc-editor.org/rfc/rfc8707.html#name-authorization-request come up when designing / looking into AutoAuth?
#
sknebel
don't think so
#
sknebel
I guess that's a more specific replacement for the realm stuff from HTTP?
[Paulo_Pinto] joined the channel
#
jacky
hmm yeah
#
jacky
that was really it tbh :)
#
jacky
I wanted to work on AutoAuth today but I'm still working on fixing async Webmentions for my ite so if I get to it, I'll mention it
#
aaronpk
ok first project is done, now to decide between autoauth or micropub extensions
#
GWG
aaronpk: I respect either decision
#
aaronpk
sknebel: jacky: do you want to talk about autoauth for a bit?
#
jacky
I'm still open to it
#
jacky
like I want to definitely get some more contextualization in it; I'm preparing to work on Fortress again
#
aaronpk
ok let's grab a zoom room, one sec
[asuh] joined the channel
#
aaronpk
zoom link is in the autoauth page on etherpad
#
GWG
It is?
#
GWG
That page shows two archived sessions
#
aaronpk
* autoauth section on the projects etherpad
#
GWG
Going to listen in
nickodd left the channel
#
[tantek]
snarfed, another reason for manual first (along the lines of manual until it hurts), at least for multi-person features: manual first to better understand e.g. social dynamics, especially power disparities, privilege, respect, safety, mutual agency
#
[tantek]
abstracted from our person-tag session yesterday
#
[tantek]
What is manual until it hurts
#
Loqi
manual until it hurts is an indieweb principle of resisting automation until you have done it enough times to really understand it, and know that it is worth doing https://indieweb.org/manual_until_it_hurts
#
[tantek]
Yup, need to expand and/or perhaps even rename that accordingly
#
[tantek]
Kinda meta-dev so it seems worth keeping here. Methodology of building things, making things easier, lowering barriers, automating etc
mjordancodes and [jgmac1106] joined the channel
#
[jgmac1106]
Really like the idea of micropub halfpoint we discussed last night. Give a copy/paste option before you login with IndieAuth.
[mJordan] and chrisaldrich joined the channel
#
[jgmac1106]
Still a barrier as you are using plain HTML but lower barrier than supporting a micropub endpoint on a website
[Chaitanya] and Hibs joined the channel
#
sivy
[tantek]: “]: sivy, just wait til you have to backup or migrate, then every other storage system tends to break” - well that’s why I’m stuck right now, I want the FS storage but what to do with structured webmention data… ~grr~
leg joined the channel
#
sknebel
sivy: what's the specific problem with putting the webmention data in files?
#
[tantek]
sknebel, how to
#
[tantek]
sivy, I 100% sympathize, I'm literally stuck on that too, so I'm a good person to ask questions to/with, however not having solved it personally, I'm not a good source of "how to"
#
[tantek]
sivy, then again, I'm looking it from a "how do I process 200,000 inbound webmentions into some form of scalable storage" perspective
#
[tantek]
and that's just from Bridgy
#
[tantek]
for example, I'm not convinced that I want to or should try to store 100% of all webmentions's raw contents.
#
sknebel
yeah. I do that - just a file per webmention received with the data dumped in. for rendering it I'd probably want some form of postprocessed version (e.g. sivy's database as a cache), and your 200k are ofcourse a lot more than I process
#
[tantek]
e.g. for some I might only store the "log" data of "a webmention happened at this time, with this from/to, and was / was not / unknown if verified at that time"
#
[tantek]
lol yeah I'm not going to create 200k files 🙂
[Paulo_Pinto] joined the channel
#
[tantek]
I'm not even sure I want to store even just the *occurrences* of all webmentions. I may instead store blockranges for some URLs/paths / profiles where I don't even know which webmentions if any were received or not, just that I may have rejected/ignored them.
#
[tantek]
E.g. if you block someone, you don't necessarily want to care about keeping track of every single webmention they've sent you
#
sknebel
yeah, if you reject them you likely won't need all those details
geoffo joined the channel
#
sknebel
and no need to record them at all, or just in some really simple log somewhere *maybe*
#
[tantek]
however you may need to keep track of (in storage) the fact that you did reject them, or have been rejecting them for a period of time
#
[tantek]
right, simple log is a good fallback from full storage
#
[tantek]
yet even there, if someone is being abusive and sending spam or or worse I may not even want a persistent log of their permalinks myself (like I don't want to accidentally view/click them in the future or even have it as a trap for someone else)
#
[tantek]
so in that case a way to "mask" such abusive webmentions would be to keep perhaps just the domain / profile URL and a start date for when blocking started, perhaps an optional "why" (plain text or a URL of the post that made me block them), and optional blocking end date.
#
[tantek]
next level up from that is blocking whole *networks* rather than having to keep track of individual people / accounts / profiles
#
[tantek]
e.g. perhaps even pre-block abuse pits like G*b
#
[tantek]
so really, to solve webmentions storage, I have to first solve network/site-level blocking 😭
#
petermolnar
(swearword incoming) - regex filters?
#
[jgmac1106]
[tantek] gonna share a minimal storage in a spreadhseet you get with Martin Hawksey's Twitter Archive for Google Sheets....what if it was one file per post...with tablature data for webmentions
#
[jgmac1106]
Darius did some fun stuff I think with storing webmentions in a spreadsheet
#
[tantek]
[jgmac1106] my storage problems are far above the plumbing details lol
#
[jgmac1106]
yeah 200k.....
#
[tantek]
please re-read the above
#
[tantek]
petermolnar - that's a good point, might need to consider blocked words/hashtags too
#
[jgmac1106]
just wondering if one file with x number of records is better than x number of files from a noob perspective
#
[tantek]
[jgmac1106] ah, that I have an answer to. file with a number of records is better *if* you can make it algorithmically predictable how to find which file has which record.
#
[tantek]
e.g. my bim storage files
#
[tantek]
it may require slightly more code, however has big storage scaling advantages
#
[tantek]
e.g. this is why I have one file per bim, rather than one file per post for my own content
#
[jgmac1106]
then deciding display: first time you sent me a webmention, someone I follow, 2nd degree connection.....goes to learn what a bim is
#
[jgmac1106]
What is bim?
#
Loqi
bim is short for newcal bi-month, or 61 day period (except for the last 60 days of a non-leap-year) https://indieweb.org/bim
#
[jgmac1106]
What is newcal?
#
sknebel
aaronpk: jacky: there's also this issue re splititng out pieces: https://github.com/sknebel/AutoAuth/issues/17 (in addition to splitting out the pieces of the authorization endpoint)
#
Loqi
[sebsel] #17 Should the token endpoint be shared with normal IndieAuth?
#
[jgmac1106]
yess...it gets back to 6 base number system.....but got to stop this rabbit hole
#
[jgmac1106]
only here can you get from how should I store webmentions? to we need to rethink our entire calendar
#
petermolnar
btw 200k files in a single directory, while it sounds horrible, won't break any modern fs
[scottgruber] joined the channel
#
sknebel
+ you can always bin by date etc
#
sknebel
but database as cache probably makes sense for searching through that in flexible ways
#
jacky
still has yet to implement search
#
[jgmac1106]
^ just posses to twitter and uses Google to search his website because search is no longer effective
#
Loqi
Lighthouse
#
jacky
I noticed it to be a bug when looking at how sending Webmentions should work
gRegorLove and geoffo joined the channel
#
[Chaitanya]
How can I know SHA of a Github project/file for Integromat? Thank you!
[fluffy] joined the channel
#
[fluffy]
By SHA do you mean commit hash? that’s the long hex string that’s associated with each commit.
#
[fluffy]
so for example the latest commit of https://github.com/PlaidWeb/Publ/ has a hash of 1e83cb09bbd1c2ae88b481dd4e51a909e27fca7c which you can get by clicking on the “latest commit” link (or by checking the repo out and doing `git rev-parse HEAD` or the like)
#
[fluffy]
I don’t see an easy way to get a file’s object hash from github although there probably is one. But you usually don’t need those.
#
dansup
I wonder what the best way to markup vlog posts would be
#
dansup
I want to make my own design and support AP and indieweb so it will be a new laravel project, can't wait :D
#
jacky
dansup: _expand_
#
jacky
this is on my to-do (from a quick search)
#
jacky
like a weekly vlog
#
jacky
(more like every two weeks)
#
dansup
yeah, I just wanted to use something other than mastodon
#
jacky
tbh if it's just like [video, text] then that's a /video post up front
#
jacky
the design's up to you!
#
[fluffy]
What’s the name of the ActivityPub YouTube clone?
#
sknebel
peertube?
#
[fluffy]
I have an account on one of them although I haven’t used it much. The whole platform seemed to have been overrun by alt-right vloggers.
#
[Chaitanya]
[fluffy] Maybe it is something different. That hash from my repo didn't work.
#
jacky
yeah it's peertube
#
jacky
they still haven't pushed any concept of moderation :(
#
[fluffy]
[Chaitanya] hmm, I dunno what it might be looking for then. I haven’t tried Integromat yet so I don’t know what its GitHub integration is like.
#
dansup
jacky: yeah its just video and a bit of text, the videos are pretty short https://mastodon.online/@gansup/104418012928318134
#
Loqi
[gay dansup] Hey cuties and fedi devs!This is my first vlog 🥳 [video]
#
[fluffy]
I would think it only needs to be a thing you can assign a GitHub webhook to.
#
[fluffy]
I guess it depends on what sort of integration you’re doing though.
#
[Chaitanya]
I am trying to update a json file on Github with weather data.
#
jacky
dansup: this is also the first time I've heard your voice lol
#
dansup
lol
[grantcodes] joined the channel
#
[Chaitanya]
[fluffy] I did it! yayaya
#
Loqi
woot
#
aaronpk
[tantek]: got a sec to come into the autoauth room? got some questions about the UI aspect of this reverse-follow-request things
gxt_ joined the channel
#
[fluffy]
[Chaitanya] Cool, what did you end up doing? Might be a good reference for others, especially if you write this up on the wiki
#
aaronpk
zoom link is in the autoauth section here https://etherpad.indieweb.org/2020-West-Projects
#
[fluffy]
well I guess I should join that room too 🙂
#
[tantek]
scrolls
#
GWG
aaronpk: How's autoauth going?
#
jacky
it's def going
#
sknebel
GWG: aaronpk made a working autoauth implementation, we looked at some other protocol things, now he and tantek and fluffy are discussing follow/share request UX patterns etc
#
GWG
sknebel: Okay... I went to take a break, get some food, and check the windows for leaks
#
jacky
what is invitation spam
#
Loqi
It looks like we don't have a page for "invitation spam" yet. Would you like to create it? (Or just say "invitation spam is ____", a sentence describing the term)
#
jacky
what is spam
#
Loqi
spam is unsolicited content, often transmitted via email, submitted as blog comments or posted to popular wikis https://indieweb.org/spam
#
jacky
invitation spam is spam#Follower_Spam
#
[Chaitanya]
[fluffy] I used Github Gists instead. That didn't require any additionals, but did just enough to work for me.
[KevinMarks] joined the channel
#
aaronpk
sknebel: what am I missing in this example?
#
aaronpk
on my screen
#
[KevinMarks]
joseph, I was marking up the author info inside the h-entry body
#
[chrisaldrich]
What are tools?
#
Loqi
Here are some tools that are helpful for building and debugging an IndieWeb site https://indieweb.org/tools
#
[fluffy]
This discussion about doing things other than AutoAuth is interesting but I’d still like to try AutoAuth out as an overall idea going forward. Is the simplified protocol documented anywhere, and how complicated is it to actually add those pieces to wherever it’s necessary?
#
[fluffy]
I still like the usage model and UX implications of AutoAuth.
#
[chrisaldrich]
joseph_dickson, you'll find some useful things including some other parsers in that tools link above including Kevin's unmung tool, pin13 is pretty good for that too...
#
[tantek]
what is invitation spam
#
Loqi
It looks like we don't have a page for "invitation spam" yet. Would you like to create it? (Or just say "invitation spam is ____", a sentence describing the term)
#
[KevinMarks]
so it looks like you have microformats 1 markup in the site, and that's giving the support we see now - is this a wordpress theme?
#
GWG
[KevinMarks]: Easy enough to fix.
#
[KevinMarks]
I defer to GWG on the best way to fix wordpress markup
#
GWG
I yield my time to the gentleman from Yorkshire
#
[tantek]
invitation spam is a particular form of [[spam]] that has the form of offering to the recipient, access to a [[post]] or [[stream]], membership in a [[group]], participation in a [[event]], or some other capability that is typically not available to the public or just anyone, though there are exceptions like being invited to [[like]] a [[Facebook]] page.
#
[tantek]
what is a guest pass
#
Loqi
It looks like we don't have a page for "guest pass" yet. Would you like to create it? (Or just say "guest pass is ____", a sentence describing the term)
#
[KevinMarks]
I've had more literal invitation spam, thanks to gmail turning emailed ics into calendar entries
#
[tantek]
guest pass is a feature of at least one silo ([[Flickr]]) that allows a user to create a special URL to an otherwise private or protected [[post]] or [[collection]], that includes a code to access that thing, such that the user can send that URL to others without a silo account, like over email or [[chat]], allowing them to view that thing without signing in or otherwise creating a [[silo]] account.
rebelle joined the channel
#
[tantek]
both those dfns are because we started talking about such things in the AutoAuth chat, in the Red Room if you're interested [KevinMarks]! you should join us! see the project etherpad for the Zoom link
#
[KevinMarks]
I have been pretty absent this weekend as it's R's birthday, but I can lurk
#
GWG
[KevinMarks]: Happy birthday, R.
#
[KevinMarks]
zoom is crashing for me on the hallway link
#
Loqi
ok, I added "https://help.flickr.com/en_us/create-or-delete-temporary-guest-passes-in-flickr-Bygpmj1Q" to a brand new "See Also" section of /guest_pass https://indieweb.org/wiki/index.php?diff=70832&oldid=70829
#
ben_thatmustbeme
i'm going to end up forking some of these php libs if we don't start actually releasing versions
#
ben_thatmustbeme
as they stand now, you have to user dev-master and get composer requirements issues galore
#
aaronpk
which ones?
#
aaronpk
you could help release some versions? :)
#
GWG
ben_thatmustbeme: I think we were close to releasing a new php-mf2.
#
GWG
I know I did some gardening there
#
[KevinMarks]
remember the needs-tags tag?
#
ben_thatmustbeme
i will, its just that packagist needs them all, and then they need to depend on pacagist released versions
[schmarty] joined the channel
#
aaronpk
i'm pretty sure all the ones I use are tagged properly
#
aaronpk
cause i use versioned dependencies in composer
#
ben_thatmustbeme
i have link-rel-parser with dev-master but then trying to include indieauth/client:dev-master gives (weirdly) errors because p3k/http requires specific versions of link-rel-parser
#
sivy
well, at this point I’m going to go forward with storing the webmentions as YAML, but I think one yaml per target-post?
#
ben_thatmustbeme
trying a different order of them to figure out where the problem started
#
sivy
[tantek] sknebel — so if a target post is webmentioned >1 time, there would be a yaml/json file with all the webmentions for that post
#
aaronpk
ben_thatmustbeme: i think it's just a matter of finding the right mix of versions to use. it's definitely possible that the latest version of all of them aren't compatible with each other
#
[KevinMarks]
do you have moderation flags in the yaml?
#
sivy
what are the cases where one might get a bridgy webmention?
#
Loqi
It looks like we don't have a page for "cases where one might get a bridgy webmention" yet. Would you like to create it? (Or just say "cases where one might get a bridgy webmention is ____", a sentence describing the term)
#
sivy
I had not thought about 3rd party bridging, assumed I’d have minimal first-party webmentions
#
sivy
[KevinMarks]: yeah, there’s an ApprovedState
#
sknebel
sivy: I'd do something where you can just append it as text to a file then - i.e. not one large json object, but multiple ones
#
[KevinMarks]
if you tweet your post link out, bridgy will send tweet replies as webmentions to that post
#
ben_thatmustbeme
indieweb/comments failed for me
#
sknebel
my implementation does one file per webmention received, many files per post
#
sivy
I think there would end up being a “post-data” table that maps those to posts, and lets me create a listing of webmentions to approve/reject
#
ben_thatmustbeme
mf2/shim failed too
#
sivy
this is where my confusion on how to store this stuff comes in — the DB is easy: targetUrl, sourceUrl, rawhtml, <other optional bits>
gRegorLove joined the channel
#
ben_thatmustbeme
though i think i saw you have a p3k/twitter-shim
#
sivy
but mapping that to an easy-to-rebuild-from filsystem representation is frustrating me
#
ben_thatmustbeme
or something like that
#
sknebel
one yaml file per line, with those values as properties?
#
aaronpk
the "shim" stuff hasn't been touched in a long time. I've been using XRay now
#
sknebel
or a file per post with one json object per line with those properties?
#
sivy
yeah, i wondered about target-post.yaml, containing multiple yaml documents, one per mention
#
[KevinMarks]
in practice, the webemntion tools tend to store parsed mf2 as well as raw html
#
sivy
it’s possible to get fs contention if lots of traffic comes in
#
sivy
“parsed mf2” - ie MF json?
#
[KevinMarks]
or further processed into reply types
#
sknebel
If I remember right, mine does a file named <timestamp><source url><randomness>.yml with some parsed data and the first X kB of the html for each webmention
#
sivy
I have implemented PTD, but have not decided on storing anything particularly structired about it right now. down such a rabbit hole, I haven’t even eploye basic webmention support yet
#
sknebel
since I don't display WMs, I don't really use the stored data tbh. I get my notifications and click the link
#
sivy
ah, i definitely want to display them
#
sivy
i really miss the old web :D
#
sknebel
right. if I did, I suspect I'd want a database cache too. but I collect hte data for now, for when/if I decide to display them
#
[fluffy]
BTW dreamwidth uses the same access model as livejournal and they have a brief explanation of it at https://www.dreamwidth.org/support/faqbrowse?faqid=124 although that’s mostly in relation to how it differs to lj
[David_Bryant] joined the channel
#
[fluffy]
but I think you can read between the lines to get a good picture of how it works
xsteadfastx joined the channel
#
[fluffy]
lj did actually have the access groups they describe but it wasn’t automatically set up and I think free users only got one or two groups, dw gives it to everyone
#
sivy
the draw of social media is obviously the feedback loop, and blog comments are just a tarpit now
#
[tantek]
yeah. though tarpit vs trash fires idk
#
sivy
trashfire works too
#
sivy
:)
#
sivy
with webmentions, my blog engine has grown beyond what
#
[fluffy]
sometimes briar patch is a better metaphor but that might have racist undertones given its origin
#
sivy
’s easy to keep in my head
#
sivy
so progress has slowed
#
sivy
feeling like i need to round-trip back to the storage layer and clean some things up. right now I write twice - once to the FS and once to the DB
#
sivy
I think i’m going to simplify so that content is *always* written to the FS first, with an immediate re-parse into the DB
#
[KevinMarks]
that is a good approach. being able to reparse later when parsers improve is good
#
sivy
yeah, and it means I keep the original content in git, and can blow away the server, rebuild from ansible and re-index the content
#
sivy
if required
#
[KevinMarks]
not being able to reparse without recrawling was a pain at technorati
#
sivy
course or that to work i need to automate the content push back to github
#
[KevinMarks]
[schmarty] I tried some stuff on glitch before at https://it-me-web.glitch.me/ - I didn't manage to work out how to make it easily cloned though
#
[schmarty]
[KevinMarks] i'm hoping that by offloading the "actual" site to neocities i can be more opinionated about how the glitch app works, offer more features and walkthrough-like features with fewer things to understand and configure up front
#
ben_thatmustbeme
hmm, got it figured out, just need to install them all in the correct order, and a few need to be dev-master
#
ben_thatmustbeme
one of the worst is php-comments uses dev-master of cassis
#
ben_thatmustbeme
s/uses/requires/