#dev 2018-02-09

2018-02-09 UTC
KartikPrabhu, chrisaldrich and snarfed joined the channel
#
www.chrisbolas.com
edited /IRC_People (+128) "/* Nicknames */"
(view diff)
tantek joined the channel
#
gRegorLove
snarfed: Seems something odd is happening with Granary and my quote post: https://granary.io/url?input=html&output=atom&url=http://gregorlove.com/notes/ search "Wendy"
#
Loqi
[gRegor Morrill] “Our level of discomfort with others’ transparency is reflective of our discomfort with our own shortcomings.” — Wendy Holland, Mission Gathering sermon on February 4, 2018
#
gRegorLove
Duplicates the blockquote
snarfed and [sdepolo] joined the channel
#
@miklb
Been in a dev rut, so to break out going to see how quickly I can go from 0 to live #indieweb site sending/receiving webmentions with @eleven_ty. Since it supports liquid, I can use my work from jekyll-indieweb as a heads start https://github.com/miklb/jekyll-indieweb
(twitter.com/_/status/961783010604605441)
[miklb] joined the channel
#
tantek.com
edited /Leaders (-90) "that was an odd Loqi bug to fix, concerning that I somehow snuck a string into the static "See Also" heading that Loqi added"
(view diff)
tantek and renem joined the channel
#
tantek
!tell aaronpk interesting that your post permalink https://aaronparecki.com/2018/02/08/1/ doesn't show your own follow-up replies to my comment! I'd expect to see the whole thread there.
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
[Aaron Parecki] Portland to San Francisco https://aaronparecki.com/2018/02/08/1/map.png
chrisaldrich joined the channel
#
www.boffosocko.com
edited /Medium (+472) "rel-canonical setting for WordPress plugin; pagelogo; moved TOC"
(view diff)
[colinwalker] joined the channel
#
aaronpk
That would be nice!
#
Loqi
aaronpk: tantek left you a message 1 hour, 4 minutes ago: interesting that your post permalink https://aaronparecki.com/2018/02/08/1/ doesn't show your own follow-up replies to my comment! I'd expect to see the whole thread there.
#
aaronpk
It'll happen when I implement threading. There's nothing special about my own replies tho
[eddie] joined the channel
#
[eddie]
The worst part about iOS development? Code signing and device provisioning. The latest build of Indigenous (full of bug fixes) works fine in my device and simulator but fails when building for TestFlight
#
aaronpk
It's gotten so much better too
AngeloGladding, KartikPrabhu and [jeremycherfas] joined the channel
#
[jeremycherfas]
[eddie] For the record, Quill has never been able to find my syndication targets from Known.
barpthewire, tantek, cweiske, snarfed, [gerwitz], KevinMarks and [kevinmarks] joined the channel
#
[kevinmarks]
It used to be able to. The version on my phone works, but when I connect again it doesn't work any more.
tantek and snarfed joined the channel
#
[12 lines deleted]
# jeremycherfas joined the channel
[1 line deleted]
#
tantek
zegnat wat
#
jeremycherfas
That phony760 is so weird. I happen to have Slack open at the same time, and there the message shows as a bunch of black blocks and double border thingies.
#
jeremycherfas
Anyway, somebody block it, please.
#
tantek
jeremycherfas: banned it
#
jeremycherfas
Thanks tantek
#
Zegnat
Should probably remove that from the logs again.
#
tantek
jeremycherfas: want to help with removing it from the logs?
#
jeremycherfas
Don't know how.
#
tantek
what is removing irc spam
#
Loqi
It looks like we don't have a page for "removing irc spam" yet. Would you like to create it? (Or just say "removing irc spam is ____", a sentence describing the term)
#
tantek
removing irc spam is /IRC#How-to_remove_spam
#
loqi.me
created /removing_irc_spam (+35) "prompted by tantek and redirect added by tantek"
(view diff)
#
Zegnat
I’ll have a PR ready to remove this in like 5 minutes
#
tantek
cweiske thanks for the image. any idea why a bot would advertise just another IRC server?
#
cweiske
to get people to connect to it, as spammers usually do
#
Zegnat
Ah, can’t do the PR yet, it hasn’t been pushed to the GitHub logs yet.
#
cweiske
to grow their server or so
#
jeremycherfas
Zegnat It would make life even easier if https://indieweb.org/IRC#How-to_remove_spam contained a link to where to find the logs in github.
#
jeremycherfas
Not the day, but the "folder"
#
Zegnat
The Spam section has a link to the GitHub archive, I didn’t duplicate that link again in the section I wrote. Maybe should have. I’ll do that now.
#
cweiske.de
edited /IRC (+73) "/* How-to remove spam */ link github"
(view diff)
#
Zegnat
Ah, crap, I was editting to so now I have a conflict with cweiske’s edit
#
vanderven.se martijn
edited /IRC (+174) "/* How-to remove spam */ Put the edit screenshot in the right list item, add information on where to find the files."
(view diff)
#
Zegnat
PR submitted
calumryan joined the channel
#
Zegnat
jeremycherfas, please see if the How-to on spam is more clear now :) I originally wrote it trying to create a step-by-step guide any one could follow
#
Zegnat
(though it still depends on having a regular expression compatible text editor :( )
#
@sl007
@OKNRW @derarndt Seit 4 Jahren arbeite ich an einem dezentralisierten CMS – einige der Technologien (Micropub, Webmention, ActivityPub) sind inzwischen W3C-Empfehlungen. Also nicht falsch verstehen, Eure Idee ist toll. Aber BITTE überlasst das Digi-Feld bundesweit nicht auch noch den Rechten.
(twitter.com/_/status/961898047549661184)
#
jeremycherfas
That's great for me, and I do know what I'm doing with BBEdit! So I'm armed and dangerous.
#
Zegnat
I really like BBEdit, one day I might buy it. I’m on the evaluation edition.
[kevinmarks] joined the channel
#
jeremycherfas
I bought it a long time ago and it is almost certainly overkill for me, but I really like things like multi-folder grep.
snarfed joined the channel
#
[kevinmarks]
Also the folder diff is very good
#
jeremycherfas
And it is very good about working over SFTP
#
sknebel
[kevinmarks]: what do you mean with oembed "html variant"?
#
@feross
🙌 Just released a CLI tool called `thanks` to help you thank the open source maintainers you depend on! ✨ 1. Run 'npx thanks' in your project 2. See which of your dependencies are seeking donations! 💸 🌟 Open source authors, add yourself to the list: https://github.com/feross/thanks https://pbs.twimg.com/tweet_video_thumb/DVjOet0VQAA2CYg.jpg
(twitter.com/_/status/961745970080792576)
#
[kevinmarks]
The rich type of response
#
sknebel
i *think* wordpress does that, would have to double-check though
#
Zegnat
I don’t know if anyone supports it. I don’t usually check support before implementing something I am interested in offering.
#
Zegnat
But it feels like a good way to enable reposts.
KevinMarks joined the channel
#
www.svenknebel.de
edited /oEmbed (+1) "/* Discussion */ fix broken fragment for IRC logs"
(view diff)
#
sknebel
Zegnat: add a note about that to ^^^
#
vanderven.se martijn
edited /oEmbed (+406) "/* Discussion */ oEmbed for reposting"
(view diff)
#
Zegnat
Done sknebel
#
Zegnat
I’m still wondering if there is a better way to do it. But oEmbed being an established standard is a bit of a plug
#
Zegnat
s/plug/plus/
cweiske, snarfed and calumryan joined the channel
#
dgold
okay, I have aperture and watchtower functioning
#
Zegnat
Nice!
#
Zegnat
Did you happen to write up a guide, dgold? ;)
#
dgold
I shall be doing to imminently
#
dgold
hmmmmm
#
dgold
indigenous isn't seeing my feeds at all.
#
tripu.info
created /User:Tripu.info (+54) "Let there be light"
(view diff)
#
jeremycherfas
Here we go again.
#
jeremycherfas
That's pretty cool dgold
#
dgold
for indigigenous, should I be setting the microsub up separately, or is the link in the micropub endpoint sufficient?
#
dgold
ah, scoping is all wrong
snarfed, barpthewire1, voxpelli, leg, KevinMarks, deathrow1 and [eddie] joined the channel
#
[eddie]
!tell jeremycherfas Interesting. Good to know! Do you have the latest Known code or an older version?
#
Loqi
Ok, I'll tell them that when I see them next
#
dgold
oh, [eddie] ...
#
jeremycherfas
I'm on version = "0.9.9" build = 2017051501
#
Loqi
jeremycherfas: [eddie] left you a message 2 minutes ago: Interesting. Good to know! Do you have the latest Known code or an older version?
#
[eddie]
Hey there, dgold! The Microsub endpoint in the header should be enough
#
[eddie]
But you should also have a read scope
#
dgold
did you have to additional wrangling/wrassling with aperture to get the scopes working?
#
dgold
I do have a read scope
#
dgold
read follow block mute channels
#
[eddie]
Nope, it just worked for me. That said it worked after I submitted two PRs so, you’ll want to be sure you have the latest code from aaronpk’s branch
#
dgold
how do I check the scope being acquired by indigenous
#
[eddie]
Hmmm that’s a good question. I think I need to add that in to the account debug screen because I don’t think there is currently a way to see that 😕
#
[eddie]
Headed out to an appt, be back in an hour or two
#
Zegnat
If the auth token is in de debug screen, you can make a request with that to your token endpoint yourself dgold. The token endpoint will tell you what scopes that token has.
eli_oat joined the channel
#
Zegnat
Looking at the aperture code, I think we can also cheat if you need some scopes in there without regenerating the token in indigenous. But probably best not to start of with a hack :P
#
dgold
Zegnat: hmmm. how do I make that request?
#
Zegnat
No, sorry, wrong link
#
Zegnat
https://indieauth.spec.indieweb.org/#access-token-verification - is what you want. Just go a GET request to your token endpoint, with the Authorization header filled with the token that Indigenous is using.
#
Zegnat
In your case `curl -H "Authorization: Bearer %%%" https://tokens.indieauth.com/token` on the command line should do the trick. just replace %%% with the actual token.
snarfed and leg joined the channel
#
jeremycherfas
Zegnat You remember that business about finding out the fields Known will accept by using a GET to the endpoint? Does it have to be an authenticated request?
#
jeremycherfas
When I try a very simple request https://stream.jeremycherfas.net/status/edit I seem to get the content of the page, not the response I am looking for.
#
jeremycherfas
I suppose I need to generate the secret details and try again with that.
snarfed1 joined the channel
#
dgold
read follow mute block channels
#
dgold
but it still gets 400 from aperture
#
Zegnat
It has to be a POST request. It can probably be done with either your current session (as you have it in the browser), or by passing the API key in a header
#
Zegnat
jeremycherfas ^^^
#
Zegnat
dgold. Hmm. Then I can’t comment I’m afraid. I don’t run Aperture my self :(
#
jeremycherfas
Zegnat, but "Many pages report some basic information about them by making GET requests, so for example, making a GET request to the same /status/edit endpoint will return, among other things, available syndication methods and the type of object being created."
#
dgold
its fine, Zegnat, I may have to wipe and reinstall - keeping the .env settings of course
#
aaronpk
dgold: that sounds strange
#
aaronpk
making breakfast but will help in a bit
#
Zegnat
jeremycherfas I never actually checked where the code is that does the whole “every page is an API” thing.
#
Zegnat
I just checked the saveDataFromInput methods on the different Entity classes to figure out what inputs were allowed.
#
jeremycherfas
No worries. I'll get there. Got to get my bread goihg.
[gerwitz] joined the channel
#
[gerwitz]
y’all talk about food in here too much. Now I’m hungry.
#
dgold
aaronpk: don't worry about it - I reinstalled
#
dgold
this time it brought in a whole lot more dependencies, now it functions.
#
dgold
I appear to have done 'something' to my gnutls package at some point, I had to fix that first
#
Zegnat
[gerwitz], go subscribe to jeremycherfas’ food podcast. It is good ;)
#
dgold
+1 to that suggestion
#
dgold
its as good (in its way) as 99pi
#
aaronpk
oh funny
#
[gerwitz]
(If it doesn’t have “veg” in the title, it’s probably not gerwitz-compatible)
#
dgold
[gerwitz]: jeremy's 'cast does veggie. and bread
#
dgold
he had a great one a little while ago about US Food Policy in the 70s and 80s, got to speak to some of the architects of the policy too
#
jeremycherfas
The hard thing about working on Known's "each page is an endpoint" thing is that each type of page requires a different signature for authentication. Which is a real pain.
#
jeremycherfas
!tell cweiske should I uninstall my older version of shpub before getting the new one?
#
Loqi
Ok, I'll tell them that when I see them next
#
Zegnat
Alright. I should have left a review now. I think. My iTunes is acting real weird :/
#
Zegnat
woops, wrong channel
#
dgold
jeremycherfas: you can just delete the binary
#
dgold
where are you installing it to?
#
dgold
if its on your path, just copy the new one to the same location, it'll overwrite the old
#
jeremycherfas
It has a different version number.
#
dgold
just $ cp shpub-1.2.3.php ~/bin/shpub
#
dgold
you should strip the version number in use
#
jeremycherfas
And weirdly, the commands that used to work with the old version, like shpub-0.6.0.phar server now give an error "shpub-0.6.0.phar: command not found"
#
Loqi
it is probable
#
jeremycherfas
Currently it isn't in a directory. I can try that.
#
jeremycherfas
Cp rather than mv?
#
dgold
mv is also good, I just use cp to keep archives of older versions
#
jeremycherfas
Now I'm completely screwed. Md gives command not found. Must be something very odd.
#
dgold
what's now in your ~/bin/
#
dgold
(as in - what's the shpub in the bin)
#
jeremycherfas
Shpub-0.7.0.phar
#
jeremycherfas
But no bash commands seem to be working.
#
jeremycherfas
Well, ls is working.
#
jeremycherfas
Ah. My permssions are all screwed.
#
jeremycherfas
I am the owner of my ~/bin aren't I? So surely I can mkdir in it?
#
dgold
ok - why are you putting it in a drawer in your ~/bin
#
dgold
ls -la ~/bin
#
jeremycherfas
And you suggested the same, no?
#
dgold
but... you're not putting it in /usr/local/bin, you're putting it in ~/bin
#
jeremycherfas
Yeah, sorry. I did edit the command.
#
dgold
then just $ mv /path/to/shpub-x.y.z.phar ~/bin/shpub
#
dgold
don't put it in a _drawer_, then you'll have ~/bin/shpub/shpub-x.y.z.phar
#
jeremycherfas
What were you saying about renaming the phar file?
#
dgold
there's no need to do that if you do the $ mv command as outlined
#
dgold
@ 15:57:24
#
dgold
(my time)
KartikPrabhu and [eddie] joined the channel
#
@jacobian
Has anyone done a Django implementation of micropub and webmention?
(twitter.com/_/status/962002367074873344)
snarfed, tantek, KevinMarks and jackjamieson joined the channel
#
jackjamieson.net
edited /Yarns_Indie_Reader (-17) "Removed mention of support for micropub (Yarns does not support micropub)"
(view diff)
[eddie] joined the channel
#
[eddie]
!tell aaronpk: Does Aperture have any way to funnel posts into the notifications channel? The Microsub spec defines a home and notification channel but I’m not sure how Aperture deals with those right now
#
Loqi
Ok, I'll tell them that when I see them next
#
aaronpk
not yet
#
aaronpk
actually
#
aaronpk
now that there's a micropub endpoint, you can write posts to the notifications channel
#
aaronpk
one of my plans it to make Aperture handle home page webmentions and have those go into the notifications channel. but you could kind of hook that up manually right now if you wanted
#
[eddie]
Does the notification channel exist? Where I could Micropub to it? Or no, it doesn’t exist in Aperture yet?
#
aaronpk
it does
[miklb] joined the channel
#
[miklb]
aaronpk how did you set up your syndicate-to for Twitter to use with micropub for likes and reposts?
#
[eddie]
🙌 gotch
#
[eddie]
Gotcha*
#
aaronpk
if you make an api key from that channel you can use it as the micropub access token and the posts will be written to that channel
#
aaronpk
the micropub endpoint is just /micropub
#
aaronpk
[miklb]: my site recognizes when a post has a Twitter POSSE copy, finds the twitter URL, then passes that to silo.pub to send my copy of the like to twitter
#
aaronpk
it was a bit of code to make that work but not too bad. and silo.pub doing all the API work made it a lot easier
#
[miklb]
I hadn’t thought about silo.pub, I was looking at using the existing bridgy one in GWG’s plugin. I guess I need to look at what is returned in the micropub like/repost to know exactly what property I can check for. Thanks
#
[miklb]
this feels like it is moving from an itch to a blocker and I want to solve it once and for all
#
[eddie]
[miklb] a like/repost would return the url of the item liked or reposted. So you would want to first, check if the url is a Twitter url. If so, Bridgy/silo.pub it. Secondly, if it’s not a twitter url, parse it for microformats and check for any twitter u-syndication links, if so Bridgy/silo.pub it
#
[miklb]
[eddie]++
#
Loqi
eddie has 18 karma in this channel (36 overall)
#
[miklb]
I was on the right track on the first part, hadn’t thought of the second
#
[eddie]
Second is definitely more complicated and not as vital. But it’s an option for completeness.
#
aaronpk
that was the part I solved last month
#
aaronpk
it was annoying me that I wasn't doing it
#
[miklb]
GWG has probably already done the heavy lifting for that second part in WordPress, I’ll just need to find the function to tie into
#
snarfed
bridgy publish also checks for synd links itself, so alternatively, you could just attempt to bridgy publish everything, and let it fail on non-syndicated posts
#
[miklb]
lol. don’t tempt me
#
snarfed
feel free
#
[miklb]
thanks, I’d rather get more familiar with micropub, and particularly, your plugin for it. I might be too intimidated to submit any PRs, but we’ll see.
#
[miklb]
s/intimidated/embarrassed don’t know how to write unit tests/
#
[eddie]
[miklb] I work as a programmer full-time professionally. Over the course of the last decade, I have written less than 10 unit tests 😉 you are in good company
#
[eddie]
That said I desire to write more, but things get in the way, in hobby projects and in the everyday 9-5 world
#
snarfed
alternative perspective: i work as a programmer full time too. over the last two decades, i can probably count on one hand the things that have fundamentally changed the way i work. unit tests are at the top of that list, unequivocally for the better.
#
snarfed
fwiw! :P
#
[eddie]
snarfed++
#
Loqi
snarfed has 30 karma in this channel (338 overall)
#
snarfed
(not that i code much at work recently, but that's a different story)
#
snarfed
[eddie]: over the medium and long term, unit tests will save you significantly more time than they take to write
#
Zegnat
I have also had a hard time fitting unit tests into my workflow. Because I am often just working off of a very specific specification on the inputs and outputs already.
#
[eddie]
I think Unit Tests are amazing. If you can, do it :thumbsup: But never be ashamed that you can’t write them. They are, as Snarfed pointed out a time saving optimization
#
[eddie]
They are totally on my list of things to start doing
#
[eddie]
But Unit Tests don’t a programmer make 🙂
#
[miklb]
oh, sure. But a contributor to snarfed’s work they do
#
[eddie]
lol makes sense
#
snarfed
[eddie]: they're bigger than just a "nice to have" imho, especially for anything that matters.
#
snarfed
most things we do here are side projects, for fun, so sure, they're not "required" for most of our projects.
#
snarfed
but anything remotely production quality or scale...i'd have a hard time taking at all seriously without comprehensive automated tests.
#
snarfed
20y ago, maybe. now, no.
#
aaronpk
I find it really hard to write unit tests for larger projects, especially when there's a UI involved. but I'm a huge fan for libraries and other self-contained parts of a project.
#
snarfed
agreed! unit testing UI is still hard, selenium etc notwithstanding. but yeah, that still leaves a huge majority of code very testable.
#
[eddie]
Hopefully I’ll be able to get into it more at work, we use Angular and so part of our current programming cycle, one of our people has started up integrating the Angular testing stuff into our programming environment
#
[eddie]
So I think it’s coming
#
snarfed
great!
#
aaronpk
and yeah I have never regretted writing unit tests when I do
#
[eddie]
At guess it’s at this point in the convo (UI) that I should point out I’m a UI Engineer.
#
dgold
oh, wow, my notes on getting Watchtower & Aperture working are looong. I don't think I can get these made into blogposts
#
[miklb]
I just haven’t found the right environment and timing to learn them. Maybe if what I come up with for the micropub plugin will be my first foray. I trust snarfed to be helpful in the PR
#
snarfed
[eddie]: aha yeah, understood. still doable, but often harder
#
aaronpk
dgold: you could send a PR to the readme from them
#
snarfed
[miklb]: sure!
#
dgold
aaronpk: i think that would be the better approach.
#
aaronpk
or maybe add to a DONTREADME.md 😂
#
dgold
I'll start reading your READMEs to get idea of how you like them written
#
aaronpk
XRay is probably one of my better documented ones if you want a good example https://github.com/aaronpk/XRay
#
Loqi
[aaronpk] XRay: X-Ray returns structured data from any URL
#
Zegnat
I find I am writing a lot of code that depends on HTTP stuff. And just having to do all the Request object mocking already turns me off of writing the tests. But I’ll get over that some day.
#
dgold
:) that was one of the two that I was looking at - that and Compass
#
aaronpk
Zegnat: yeah that's harder for sure. I spent a long time doing the setup for that in XRay but it's totally worth it now.
#
Zegnat
Diactoros actually seems to come with a really nice Request serialiser and unserialiser. Now that I have done some projects with that lib, I have been thinking about writing tests again.
#
snarfed
while you all are here, feel free to beta test a new feature on https://brid.gy/ i just deployed!
#
aaronpk
whoaaaa
#
snarfed
only publish right now
#
snarfed
hopefully backfeed eventually
#
KartikPrabhu
wait what beta test?
#
KartikPrabhu
I don't see anything new
#
snarfed
look again :P
#
KartikPrabhu
has logo blindness :P
#
KartikPrabhu
what is nascar
#
Loqi
The NASCAR problem is a reference to the jumble of branding icons on websites, e.g https://indieweb.org/nascar
#
[eddie]
hmmmm I think I’m literally blind or something is fishy. lol I don’t see GitHub anywhere
#
snarfed
[eddie]: end of the first line of buttons? refresh?
#
snarfed
oh nm, sorry, let me clear a cache
#
[eddie]
I see Flickr all the way to the right.
#
dgold
no, literally, what is nascar?
#
[eddie]
There it is! Woo!!
#
[eddie]
Excited! Nice work!!
#
[eddie]
snarfed++
#
dgold
what does it do, snarfed?
#
[eddie]
GitHub integration on Bridgy Publish, allows you to post a like of a GitHub repo to your site and then syndicate that into a GitHub repo star. Or post an issue or comment on your site and copy that GitHub
[cleverdevil] joined the channel
#
Loqi
snarfed has 31 karma in this channel (339 overall)
#
[cleverdevil]
Can't wait to find an excuse to try it, LOL.
#
[cleverdevil]
Super cool 🙂
#
[eddie]
the same way Bridgy Publish works with Twitter and Facebook
#
[eddie]
Uhoh, aaronpk, close down the issue sections of your repos!
#
Loqi
nice
#
[cleverdevil]
So, basically, you create a post in-reply-to a GitHub issue on your site, and then Brid.gy will crawl and syncidate it over to GitHub?
#
[eddie]
I think you have to Webmention it the way you do Twitter and FB
#
[eddie]
I of course am just guessing based on past conversation and bridgy publish past experience
#
snarfed
yup. interactive via the UI on your bridgy user page, or via webmention, same as the rest
#
[eddie]
posts a message on my site and now waits for Jekyll to rebuild before I can test it
#
snarfed
only issues and comments right now, no stars yet, but good feature request
#
[miklb]
well, that just changed everything
#
[miklb]
snarfed++
#
Loqi
snarfed has 32 karma in this channel (340 overall)
#
[eddie]
:thumbsup: gotcha
#
snarfed
still very beta!
#
[eddie]
issues have the repo in the in-reply-to field and comments have the comment url in the in-reply-to field?
#
snarfed
s/comment url/issue or PR url/
#
[eddie]
Oh wow you already documented it!
#
[eddie]
Talk about being on it!
#
dgold
snarfed++
#
Loqi
snarfed has 33 karma in this channel (341 overall)
KevinMarks, tantek and chimo joined the channel
#
Loqi
[cleverdevil] While I think a Gallery object would be nice, eventually, I am not convinced that its necessarily the best way to go here. Its my understanding that pretty much all Entities support attachments, so doing it in the near-term in a more cross-cutting wa...
#
[cleverdevil]
Nicely done, snarfed 🙂
#
snarfed
thanks!
#
Loqi
[Tantek Çelik] Made it to my destination safe and sound.Figured out the next big thing I want to build for my site for the #newwwyear:* POSSE issues^1 and replies to GitHubas part of my #indieweb 2018 stretch goal of owning/POSSEing all* my public posts, specifical...
#
tantek
uh oh
maingo and KevinMarks joined the channel
#
loqi.me
created /FOSDEM_2018 (+24) "prompted by tantek and redirect added by tantek"
(view diff)
[stefp] joined the channel
#
tantek.com
edited /IWS (+0) "2018"
(view diff)
#
tantek.com
edited /Summit (+0) "2018"
(view diff)
AngeloGladding, snarfed and [jjdelc] joined the channel
#
tantek.com
edited /next-iwc (-10) "until something sooner"
(view diff)
KevinMarks joined the channel
#
tantek.com
edited /MediaWiki:Sidebar (+0) "next hwc"
(view diff)
#
tantek.com
edited /next-hwc (+0) "next main"
(view diff)
#
tantek
[jjdelc]: come to the [#dev] side! >:D
#
[jjdelc]
joins
#
stephenpieper.net
created /User:Stephenpieper.net (+154) "Created page with "https://stephenpieper.net I've been adding bits and pieces of Indieweb structure to my personal Worpress site. Plugins mainly but I'd like to learn more.""
(view diff)
#
[jjdelc]
I suppose since I'm making the endpoints, my protocol understanding questions are dev related
#
tantek
[jjdelc]++ :)
#
Loqi
jjdelc has 1 karma
[eddie] joined the channel
#
[jjdelc]
[aaron_pk] thanks for your help, I am now parsing the mentions more properer 🙂 https://jj.isgeek.net/2018/02/09-014847-am/
snarfed joined the channel
#
[jjdelc]
Loqi, thanks but that preview wasn't necessary for the unrelated post
#
tantek
folks here may have a tangential interest in this proposal, feel free to follow-up in #microformats: https://chat.indieweb.org/microformats/2018-02-09/1518215307772500
#
Loqi
[tantek] Proposal: we close all classic mf (AKA mf1) examples-in-wild pages as of 2017, noting that 1. They have been so successful (millions of WP blogs etc.) that no need to track individual examples anymore, and 2. Anyone adding microformats today should b...
[kevinmarks] joined the channel
#
tantek
reads up on new bridgy feature
#
tantek
ooh interesting - there's a double-auth scenario here
#
tantek
snarfed, in the OAuth prompt to Authorize Bridgy, there's a list of "Organization access" and they're all green checkmarked, except for one, which has a grey x instead next to it, and a "Request" button
#
snarfed
tantek: yup that's common. orgs can choose their default level of oauth app access
#
tantek
clicking the Request button puts up a dialog box "Send request for approval?"
#
snarfed
(doesn't really apply to this bridgy feature)
#
snarfed
heh. feel free to try it
#
tantek
"Owners will receive an email regarding your request. "
#
snarfed
that request goes to the org owners, not to bridgy
#
tantek
"By requesting this approval, you are asking owners of the ACME organization to allow this application to access private organization data and modify public organization data."
#
tantek
why is Bridgy requesting access to private organization data?
#
snarfed
huh, bridgy only asks for the public_repo scope, which afaik is only public data
#
tantek
is there a way to further constrain the OAuth scope request to not trigger that?
#
snarfed
github's language may not pay enough attention to the scope
#
tantek
yes it still shows only Public repos for all orgs
#
snarfed
the write permissions are still overly broad (commits etc), but github doesn't have any finer granularity
#
tantek
however I think this will happen for any org you have write permission to but are not a co-owner of
#
tantek
ok I took screenshots
#
tantek
going to start conservative and *not* request that extra access
#
snarfed
oh definitely
#
snarfed
again, it's pretty common, i saw the same thing
#
tantek
in the hopes that in the future we can make Bridgy somehow not request it, or get Github to fix the default
#
tantek
it's weird because I'd still like to use Bridgy to make public comments/commits to that org's repos
#
tantek
but seemingly I can't ask for just that?
#
tantek
also the button says "Authorize snarfed" lol
#
snarfed
no, i think those org permissions are for acting as the org, not acting as you
#
tantek
is this a private beta :)
#
snarfed
you can bridgy publish to any public github repo. orgs are unrelated
#
tantek
interesting
#
tantek
then this is confusing / misleading on github's part
#
snarfed
not private beta, just soft launch
#
tantek
got it
#
tantek
would you want to use the indieweb org eventually?
#
snarfed
yes. see the scopes page above. they're somewhat granular, but could be more
#
tantek
or maybe make Bridgy its own GitHub account?
#
tantek
or maybe make a Bridgy org?
#
snarfed
orgs really are unrelated to bridgy publish. afaik you can't comment as an org, only as a user
#
tantek
(on GitHub)
#
tantek
different problem
#
snarfed
no, no, that's all totally unrelated to this functionality
#
tantek
sorry
#
tantek
"Authorize snarfed" is the green button
#
tantek
should that say instead say
#
tantek
"Authorize Bridgy"
#
tantek
or "Authorize indieweb"
#
snarfed
sure ideally. yeah maybe you can do that as an org. very low priority for me.
#
tantek
(these are the kind of nitpicky UI details that jump out at me)
#
tantek
especially for a security related UX
#
snarfed
i'm sure it was a very explicit design decision on github's part
#
tantek
like we all that know you trust you
#
snarfed
emphasizing that this isn't github, emphasizing the individual people who run it, etc
#
tantek
but if we want to abstract this into trust for the service rather than an individual, it may be worth creating a separate org / account
#
tantek
also bus factor etc. :)
#
snarfed
already mitigated a bit, kylewm has push access, oauth keys, deploy access, etc
#
snarfed
(fwiw)
#
tantek
to act as you?
#
snarfed
no, to admin all the parts of bridgy
#
tantek
uh really hope you have this snarfed: https://github.com/bridgy
#
snarfed
nope not me
#
tantek
snarfed, given the recent scares about github accounts going away (people deleting) and getting replaced by new users with same username, where such accounts used to host libraries that got included in things, this is worth thinking about
#
snarfed
account created in 2011 :P
#
tantek
sooner rather than later
#
tantek
presumably you saw that right? (account deletion, recreation by another party)
#
snarfed
heh, i hadn't but i can imagine
#
snarfed
fortunately github creds are unrelated to actually *deploying* bridgy, and also usually for pushing new versions of libs to package managers
#
snarfed
does github really let people reuse usernames after their accounts are deleted?!? that's really irresponsible
#
snarfed
yeah no shit
#
snarfed
disappointing
#
tantek
now you see my concern
#
@weekstweets
This morning, the creator of go-bindata deleted their GitHub account and someone else created a new account under the same name. Another reason to pay closer attention to the provenance and security of your software supply chain. https://lnkd.in/dvWXU_4
(twitter.com/_/status/961304627596623874)
#
snarfed
again, fortunately, not a vulnerability for bridgy specifically
#
tantek
agreed, and that's not my point
#
snarfed
interesting tangent
#
tantek
my point is people are more "sensitive" about looking at anything related to Github and security right now, so "Authorize snarfed" being the big green button while the header at the top says "Authorize Bridgy" is likely to raise some red flags
#
tantek
(bad timing as it were)
#
snarfed
understood
#
snarfed
tomorrow it'll be some other scare and we'll forget about this one :P
#
tantek
that github/bridgy account has done nothing
#
tantek
since being created 2011-11-23
#
tantek
probably worth seeing if github will free it up for actual Bridgy use
#
aaronpk
For a while IndieAuth.com said "authorize aaronpk" lol
#
snarfed
lol. username reuse is bad; let's ask to reuse a username
#
aaronpk
github changed that button at some point and didn't tell anyone
#
tantek
is it *re*use if it was never actually used for anything?
#
snarfed
(this one hasn't done anything public, granted)
#
snarfed
nothing *public*
#
tantek
my suspicion is nothing period
#
snarfed
or authed into third party services
#
[kevinmarks]
Though if they have made private repos they are more profitable for github
#
snarfed
anyway. i'll probably avoid the admin tax etc of a separate org or username for bridgy, but i understand your concern!
#
tantek
ok I'll ask on your behalf
#
tantek
which yes means I'm willing to serve as another contact for that
#
snarfed
i probably still wouldn't use the bridgy username. juggling multiple github accounts sounds like a bad time, ongoing, for an iffy workaround to a narrow problem
#
snarfed
(thank you though!)
#
aaronpk
Could get it turned into an org maybe
#
aaronpk
Multiple user accounts are annoying for sure
#
snarfed
yeah, slightly better. still not sure of the admin tax though
#
aaronpk
Probably have a better argument to make to request it for an org than a user anyway
#
snarfed
first let's see if anyone actually uses this feature, and if anyone else cares about the UI nit, then maybe it'll be worth revisiting
#
snarfed
although, actually, this is totally unrelated to where the code is hosted. it's just about which github account owns the oauth app. which makes it maybe easier to admin.
#
snarfed
anyway
#
tantek
aaronpk yes org is the goal
#
tantek
that's what I'm requesting
#
tantek
request submitted
#
tantek
let's see what happens
#
[eddie]
snarfed: for what it’s worth I thought “That’s very strange that it says Authorize Snarfed... oh well it’s a Beta” if I came in from outside the community and saw it I would have been pretty confused. That said, I’m not the most security conscious so I very well might have used it anyway
#
[eddie]
But it did raise a red flag
#
tantek
GitHub << 2018-02-07 GitHub allows account deletion, recreation by another party, with library dependency: https://twitter.com/weekstweets/status/961304627596623874 and more: https://donatstudios.com/GithubsTotalSecurityFacepalm
#
@weekstweets
This morning, the creator of go-bindata deleted their GitHub account and someone else created a new account under the same name. Another reason to pay closer attention to the provenance and security of your software supply chain. https://lnkd.in/dvWXU_4
(twitter.com/_/status/961304627596623874)
#
Loqi
ok, I added "2018-02-07 GitHub allows account deletion, recreation by another party, with library dependency: https://twitter.com/weekstweets/status/961304627596623874 and more: https://donatstudios.com/GithubsTotalSecurityFacepalm" to the "See Also" section of /GitHub
#
@francesc
📢 #GOLANG WARNING go-bindata creator deleted their @github account and someone else created a new account with the same name. There's no guarantees that the new user has good intentions, so if you're using the repository make sure you verify it first! https://github.com/jteeuwen/go-bindata
(twitter.com/_/status/961249107020001280)
#
tantek
ok that's enough on that
#
tantek
new tangent
#
tantek
back to auto-linking, in particular, @-names
#
tantek
there was a question of whether that should be done publish time or view time, also clientside or serverside (which I think is orthogonal?)
#
snarfed
lol, and you're still not signed up for bridgy github 😂
#
tantek
given that most services that use @-references allow username renaming (and account deletion / recreation by another party), there is the risk (and likely real examples already out there) of autolinked @-names that no longer refer to who they used to when the authoring was done
#
tantek
(snarfed, sorry, this is what happens with depth-first traversal)
#
snarfed
ah "new tangent"
#
tantek
e.g. CASSIS auto_link does the simple/dumb thing of "just" prefixing with twitter.com/
#
tantek
whereas now I'm thinking there needs to be a *publish-time* auto-linking step which captures the *user-id* of the @-name at publish time, and stores that in the post content
#
tantek
and then the auto-linker should turn that *user-id* into an actual linked @-name at view-time (whether on client or server)
#
tantek
does anyone actually do this?
#
snarfed
assuming the silos don't also reuse the user id :P
#
tantek
snarfed, short of domain name re-use (e.g. poco domain :P), I don't think so
#
tantek
let's just say for the sake of argument that user id re-use is far less common than username re-use
#
snarfed
domain reuse actually seems more important to consider for us specifically than silo accounts
#
tantek
so that aspect of auto-linking is also broken currently
#
tantek
now my head is starting to hurt
#
tantek
because how do you know when a domain has been re-used / re-owned in an incompatible way with the past?
#
tantek
blacklist?
#
tantek
what happens if ownership is restored? e.g. upcoming.org
#
tantek
two sidetables can help differently
tomasparks joined the channel
#
tantek
for all @-names in a post, the server should discover their user-ids and store them at publish-time as @-name,user-id pairs in a sidetable, for future auto-link resolution
#
tantek
for all *links* in a post, the (publishing) server should store the publish-time archive.org URL for each (because of course the publishing server is already pinging archive.org to Save each link right? which returns the archive.org URL for it) in a sidetable
#
tantek
these are both sidetables on the post in storage
#
tantek
sidetables must be updated whenever the post is updated
#
tantek
while keeping any changes/history? e.g. if an @-name to user-id changes, that's worth warning the user about (and likely keeping the original @-name to user-id mapping)
#
tantek
and for archive.org links, just keep each version with each update
#
tantek
s/just//
#
tantek
that's enough information to auto-link to the current @-name for any given user-id captured at publish-time.
#
tantek
and if (when?) you find out a link (domain?) has broken, you add it to a deadlinks table (manual update?) which then is a clue to the auto-linker to use the most recent archive.org URL from that sidetable instead
#
snarfed
ideally you even rewrite existing links
#
tantek
existing links in notes are all written at view time anyway by the auto-linker
#
tantek
fix-up for static (non-auto-linked) hyperlinks in /articles (or other content stored as markup) is a separate problem
#
snarfed
right, anything with html
#
tantek
(but yes, likely solvable using same datastructures / storage)
#
snarfed
separate sure, but related
#
tantek
snarfed, no, not just HTML. content stored as *any* markup
#
tantek
including markdown
#
[kevinmarks]
The piece that makes this even more painful is @ mapping across posse service's
#
snarfed
sure, of course
#
tantek
markdown has the same problem
#
tantek
explicit static hyperlinking
#
tantek
at publish time
#
tantek
KevinMarks: I think the cross-service @-mapping is orthogonal and pretty sure it doesn't complexify this at all
#
[kevinmarks]
The concrete example is instagram, which will remap @'s if the instagram userid has been mapped to a different twitter id and verified by OAuth
#
tantek
"this" meaning the need to fix the auto-linking for changes in usernames over time
#
[kevinmarks]
So if I @ name without a domain, the context is messy
#
tantek
that's already true without considering the username changing / re-use problem
#
tantek
so I assert that's orthogonal
#
tantek
kevinmarks btw IG has this problem
#
tantek
the references to @-names changing over time
#
tantek
just all by itself
#
tantek
or rather half a problem
#
tantek
two examples:
benwerd joined the channel
#
tantek
1. if you @-mention a name in an IG caption/comment, that will always link to instagram.com / thatname and never be updated, regardless if the owner of "thatname" changes it to "anothername"
#
tantek
so either those @-mentions break or you end up with old captions/comments linking to new/different accounts than they did when they were written!
#
tantek
2. if you person-tag someone in an IG photo, that *does* get auto-updated
#
tantek
IG person-tags are storing their IG user-ids and then displaying (linking) the @-name at view time
#
tantek
(yes I have found specific examples of this, but with private accounts that I can't share publicly. you should be able to verify the above with a pair of accounts and publishing from one referencing the other, renaming the other)
#
tantek
so step one - anyone who is auto-linking @-names or links, get and save the user-id for each @-name at publish time, and get and save the archive.org URL for each plain text URL at publish time
#
tantek
then in the future you can do step two and improve your auto-linking code to use said storage (either always or when necessary)
#
tantek
head still hurts
#
tantek.com
edited /GitHub (+62) "move all the related things from See Also to a new issues / Account reuse is potential security issue"
(view diff)
snarfed joined the channel