#dev 2019-07-23

2019-07-23 UTC
[asuh], chrisaldrich, [jgmac1106], [Michael_Beckwit, gRegorLove, KartikPrabhu and [David_Bryant] joined the channel
#
@easyaspython
New 1.1.0 release for django-webmention, which mostly includes non-implementation changes for improved maintainership (CI/CD and test management). https://github.com/easy-as-python/django-webmention
(twitter.com/_/status/1153504396963135488)
jolvera joined the channel
#
@easyaspython
↩️ Using tox-travis you can just tell Travis CI the Pythons you need, and all the other combinatorics will just run on the Python of interest. Pretty nice. Here's my changeset for all this jazz: https://github.com/easy-as-python/django-webmention/pull/10
(twitter.com/_/status/1153505516045381632)
jolvera, KartikPrabhu and Turtle1331[m] joined the channel
#
GWG
Okay, did a bunch of h-feed research.
[Michael_Beckwit and gRegorLove joined the channel
#
GWG
Documented in /h-feed
#
GWG
It seems that a name property is very common for h-feeds.
#
GWG
The other properties that I was looking for...author, summary, photo, and whether the feed page had pagination were somewhat less
[asuh] and [tantek] joined the channel
#
Zegnat
I was surprised to see you document a proposal by me, haha
[xavierroy], ichoquo0Aigh9ie, [Michael_Beckwit and treora joined the channel
[xavierroy] and MrKapowski joined the channel
jjuran joined the channel
#
@grabapipit
↩️ @clivewalker Ah yes, webmentions would be great.
(twitter.com/_/status/1153591680555204609)
#
@clivewalker
↩️ @grabapipit @alexhardy @perchology I still think that webmentions is a good thing to have.
(twitter.com/_/status/1153599138241679360)
maxelysium and jeremych_ joined the channel
#
GWG
Zegnat, it was noted as a todo
#
GWG
I was trying to update the page overall
#
GWG
I also fixed the fact that the markup instructions were a bit garbled
[jgmac1106] joined the channel
#
[jgmac1106]
realized way I have badges set up is basically a page for each badge and then a unique url using ID for each time the badge is issued: https://edu407.glitch.me/badges.html
#
Zegnat
Seems like a valid way to do things, [jgmac1106]. As long as parsers that you want to handle it actually support parsing h-entry elements by fragment ID
#
GWG
I need to write that out as part of the algorithm
#
Zegnat
Do we actually have a start for that algorithm somewhere? A “which mf2 object to return for a given URL” algorithm?
#
[jgmac1106]
looks pretty good, but in that one I may have the webmention being sent to the wrong url there as I would want it going to the url of the evidence
#
GWG
Zegnat, I was going to write one, but where?
#
Zegnat
Not sure what the right place would be, GWG
#
Zegnat
Maybe just a new page?
#
Zegnat
[jgmac1106]: looks good, though I am confused about the url property that seems to point to the thing I would have thought this was a reply to
#
[jgmac1106]
yeah that is what I was noting, this was an early prototype. I think I fixed that by end of semester
#
[jgmac1106]
I hired an intern to build a course builder and badgemaker. Can't wait to really start playing
#
Zegnat
Will that work be open-source?
#
[jgmac1106]
well I can't tell Marquis how to license, because he is a student worker I have to pay him an unfair below market rate...So I am letting him own the code and choose the license but just give me and anyone who buys me a...blah blah..legal stuff
#
[jgmac1106]
from discussions he is going to choose an Open Source option
#
[jgmac1106]
and i am working to Open Source ReVIEW this month. For other legal reasons I had to wait until end of July of this year
#
[jgmac1106]
Using Kirby actually to build the Course Builder and Badge Maker....Unviersity paperwork taking forever and a day to get through payroll....I am excited to start
#
GWG
Zegnat, will start on the h-feed page re discovery
#
GWG
You know, I can't find an h-feed with a photo property in the wild so far
#
GWG
But the people on the h-feed page aren't exactly current
#
GWG
As in, not much active updating
#
[jgmac1106]
podcast feeds gwg
#
GWG
Yes?
#
GWG
Do you have a podcast h-feed with a photo property to add to the examples?
#
GWG
aaronpk, thank you. I should add that to the h-feed page on the wiki
#
GWG
Where your site is not an example
#
aaronpk
tell me why i should add it to my site and I may do that :)
#
aaronpk
but right now I don't see a reason to
eli_oat joined the channel
#
GWG
aaronpk, I meant the h-feed page on the wiki
#
GWG
And percolator.today
#
GWG
I am trying to make the case for explicit h-feed being preferable to implied h-feed in general
[Rose] joined the channel
#
GWG
The h-feed page on the wiki needs more current examples. I went through and a lot has changed on people's sites
#
GWG
aaronpk, do you have any ideas where we might document a...which mf2 object to return for a given url algorithm?
#
MrKapowski
I'll add my site to the examples, GWG
#
MrKapowski
I'm planning to bake h-feed into the WP theme I'm working on, so I'll be interested in if there are any recommendations come out of your review
#
[jgmac1106]
[grantcodes] as a call back to yesterday's discussion about dt-updated would there be an easy way in Glitch to use the Git checkpoints to display the last time a page was updated rather than doing it manually?
[grantcodes] joined the channel
#
[grantcodes]
I doubt an easy way
#
[grantcodes]
There might be something of use in the glitch environment variables but I'm not sure
[eddie] joined the channel
#
sknebel
GWG: is "implied h-feed" explicitly documented somewhere?
#
GWG
I think on the feed page, as opposed to the h-feed page
#
[jgmac1106]
finished the second lesson plan if any one wants to give it a technical read and make sure Iam not saying anything incorrect: https://buildyourfirstwebsitelessonplans.glitch.me/
#
Loqi
[Greg McVerry] Build Your First Website
[tantek], farrah, gregjeanmart[m] and [schmarty] joined the channel
#
MrKapowski
How do people feel about h-feed for comments/webmentions, etc, on a permalink page? I'm thinking specifically as embedded within an h-entry. Good idea? Bad idea? Irrelevant-because-nothing-supports-it idea?
#
Loqi
Brainstorming for a Comment Microformat
#
[tantek]
MrKapowski better to ask in indieweb-dev for such a specific use
#
MrKapowski
I thought I did?
KartikPrabhu joined the channel
#
aaronpk
lol yeah that is this channel
jolvera joined the channel
#
[jgmac1106]
MrKapowski I always made a comment feed in my classroom planets, encourages greater interaction would do the same with webmentions if I built using h-feeds and not rss
#
KartikPrabhu
MrKapowski: I think right now no /reader supports this afaik
#
MrKapowski
KartikPrabhu - yeah, that matches my initial experimentation. I don't think any of the parsers picked it up either
#
KartikPrabhu
I am sure parsers should pick it up
#
KartikPrabhu
the mf2py parser does parse the example on that page correcly
#
aaronpk
I'm sure he means consumers not parsers
#
KartikPrabhu
yeah, that I expected for /readers
#
aaronpk
I've been wanting to add that to Monocle
#
aaronpk
so you can click"view comment thread" on a post
#
aaronpk
Instead of leaving Monocle to view the original post
#
KartikPrabhu
aaronpk: is a nested h-feed necesary for that?
#
KartikPrabhu
Monocole could retrieve a list of "p-comment"s directly
#
MrKapowski
I meant XRay/Parse This, etc, but I just realised why they weren't working with my test page (my fault :-) )
#
MrKapowski
there's a typo in my markup. Once I get a chance to fix that I expect it'll work
#
KartikPrabhu
MrKapowski: I would suggest not using excessive markup for your theme for now. If /readers add support of this later, then you could add it to your theme then
#
MrKapowski
it's only an extra couple of classes to implement… I'm not changing the structure of anything to add it, and lean markup is one of my goals as it is :)
[manton] joined the channel
#
[manton]
[grantcodes] I noticed that the new Together no longer works with Micro.blog's Microsub endpoint ("Error loading channels"). Any tips for what might have changed? Micro.blog still works with Monocle.
[grantcodes] joined the channel
#
[grantcodes]
Hmm maybe the same thing davidbeal has been experiencing
#
[grantcodes]
If you open up the console, there should be a link to open an issue with all the details
#
[grantcodes]
Looking in the database, it has your token and endpoints stored. (although not necessarily correct)
[miklb], [tonz] and gRegorLove joined the channel
#
aaronpk
KartikPrabhu: yes of course, I was not suggesting any particular markup
[tantek], leg and [davidmead] joined the channel
[jeremycherfas], [schmarty], dietricha and [grantcodes] joined the channel
#
[grantcodes]
!tell [manton] Figured out the microsub issue. The data fetching I am using adds a trailing slash onto all urls by default for some reason. And your endpoint returns null with a slash on the end
#
Loqi
Ok, I'll tell them that when I see them next
#
[grantcodes]
Fix is on it's way!
[KevinMarks] joined the channel
#
[KevinMarks]
What is two factor?
#
Loqi
multi-factor authentication (commonly two-factor authentication, abbreviated 2FA or TFA) is an authentication process that requires at least a second component to a standard authentication method https://indieweb.org/two-factor
[snarfed] joined the channel
#
Loqi
ok, I added "https://twitter.com/scalzi/status/1153754441562173440?s=20" to the "See Also" section of /multi-factor_authentication https://indieweb.org/wiki/index.php?diff=63877&oldid=55521
#
[grantcodes]
Random thought about your privateish pages [snarfed]: have you thought about webmentions and microformats? For example if I like it my site is going to automatically grab a copy of any mf2 on your hidden pages
[cleverdevil] joined the channel
#
@tomwiththeweath
↩️ @simondlr RSS +1 also, indieweb readers (indigenous, monocle, together, etc) can like or reply to posts that support webmention
(twitter.com/_/status/1153759799915536385)
#
[snarfed]
grantcodes sure. i'm trying to implement privacy and unlistedness, _not_ "security," ie not preventing other people from individually generating partial copies like that
#
[snarfed]
having said that, likes of private posts would do the same thing, right?
[fluffy] joined the channel
#
[fluffy]
Unlisted posts are fairly straightforward to support webmentions from, although sending a mention from an unlisted page to a public page then makes the unlisted page de-facto public.
#
[fluffy]
They’re just now listed on someone else’s site that you can’t control.
#
[snarfed]
right. grant meant webmentions _to_ unlisted pages.
#
[fluffy]
I think that’d be the same problem in reverse though
#
[fluffy]
Unlisted mentioning unlisted would be fine, any other combination would mean that the unlisted page is now listed somewhere.
#
[snarfed]
i should stop advertising my wm endpoint on unlisted pages
#
[snarfed]
also one of my goals of the "this is a private page" visible indicator was to discourage people from publicly linking to them
#
[fluffy]
Ooh yeah I should probably add an “unlisted” indicator to my unlisted pages. My actually-private pages do have an indicator though.
#
Loqi
totally
#
[fluffy]
🔏 is the emoji I use fort hat
[eddie] joined the channel
#
[eddie]
I think one thing to help with unlisted and private pages would be adopting p-visibility as a possible mf2 attribute. Something I'm planning to experiment with once I've got the basics of my iOS apps. That would allow me to basically make any replies to posts that are unlisted or private into private posts
#
[snarfed]
huh. that implies writing is restricted. wonder if there's an alternative with an eyeball instead of pen
#
[fluffy]
I don’t know why emoji has a “closed padlock with an ink pen” emoji but it seems like it’s specifically for this use case 🙂
#
[snarfed]
apologizes and slowly backs away from design
#
[fluffy]
for unlisted-but-hidden I’d probably use 🔓 (open padlock)
#
[fluffy]
although it’s kind of a subtle distinction, at least on the macOS font.
#
[fluffy]
between 🔒and 🔓anyway
#
[fluffy]
also I can’t believe I”m about to complain about bad kerning on emoji but: holy hell that’s some bad kerning
#
[fluffy]
on the macOS emoji font it’s hard to tell if there’s a space after it or not, but adding an explicit space means there’s *too much* space
#
[fluffy]
anyway I’d love to see a `p-visibility` or similar in addition to visual cues. Make it so that micropub et al clients can at least present a warning like “hey are you sure you want to do that?”
#
[eddie]
Yep, exactly. I think with unlisted and private posts becoming a key part of people's set ups over the next year, I think that'll be needed
#
[fluffy]
also I’m trying to figure out how I can reasonably receive webmentions on my actually-private posts. Right now if someone’s unauthorized they get redirected to another page to actually handle the login.
#
[eddie]
So I'm gonna add it to my stuff when I have the time and we'll see if others follow suit
#
[eddie]
Yeah webmentions to private posts is difficult currently
#
[fluffy]
So the webmention endpoint would probably do the same unless it’s authorized via AutoAuth. and I don’t like the idea of giving a potentially third-party service read access to everything.
#
[eddie]
Hmmm yeah that's tricky
#
[eddie]
I think AutoAuth has the option of a "realm" which I don't fully understand
#
[eddie]
but maybe you could separate posts into different realms? or even have a realm per post?
#
[fluffy]
I suppose I could also make it so the initial page access doesn’t redirect and just publishes the webmention endpoint for the actual target page but that gets tricky too
#
[eddie]
then the token for that realm would apply to a subset of your posts
#
[fluffy]
I’m not sure what a “realm” is in this context
#
[fluffy]
oh I see
#
[fluffy]
feels too complicated
#
[fluffy]
I mean personally I’m not a fan of other peoples’ replies to my private stuff being potentially broadcast on the public Internet anyway
#
[fluffy]
I’d much rather people use my own native comments (which are disqus at present but that’s beside the point) to reply to things that I have locked down
#
[eddie]
If you don't want replies to private stuff I'd remove the webmention endpoint on private posts
#
[eddie]
I want people to reply from their own site on private stuff, I just want to ensure they make them private posts as well
#
[fluffy]
yeah I guess I could do that, the templates could just be like `{% if not entry.is_private %}<link rel="webmention" ...>{% endif %}`
#
[fluffy]
(that being Publ-specific of course)
#
[eddie]
Yeah that would work.
#
[fluffy]
of course rmeoving the endpoint still doesn’t prevent people from posting about/linking to the post publicly
#
[eddie]
Correct
#
[fluffy]
which is the actual security concern
#
[eddie]
You can never prevent that
#
[eddie]
it's the internets
#
[fluffy]
so I mean all that removing the endpoint does is makes it less likely I find out about it 😛
#
[eddie]
True, although if you don't set up something like AutoAuth the webmention endpoint won't ever work on a private post
#
[fluffy]
I think this is a problem for future fluffy to think about.
#
[fluffy]
I hope that this becomes a problem I have to address in the future! But for now things are too scattered/vague for me to see any reason to worry about engineering a fix.
#
[eddie]
for sure
#
[fluffy]
as it stands, all of my privacy-post-enabled friends are either using an RSS reader or manually polling my site for updates, and the ones with a reader just see a placeholder stub like “🔏 Private entry [Jhs]” which is the compromise I decided on.
#
[fluffy]
basically the most recent unauthorized entry gets a stub entry with a login link and the initials of the actual post title.
#
[fluffy]
If someone happens to use a cookie jar or similar to share their login session cookie with their feed reader, they’ll get the full post.
#
[fluffy]
ooh and I just realized I don’t have the privacy indicator in the title in that case
#
[fluffy]
not that I have any evidence that anyone’s actually gotten that working though.
#
[fluffy]
this is a thing I’m hoping AutoAuth makes possible in the future but that requires soooo much stuff to happen on the reader side before it will.
#
[fluffy]
nice, I just verified that a cookie jar WILL work with a feed reader, by giving it my admin session’s cookie.
#
[fluffy]
(after adding the privacy marker)
DenSchub, releasetheghosts and [pfefferle] joined the channel
Vermillion6861[m left the channel
#
GWG
[eddie] I was looking at your website
#
GWG
Why not mark up the homepage as several partial h-feeds?
[jgmac1106] joined the channel
#
GWG
[tantek]: Would you have an opinion on the question I asked this morning?
#
GWG
Or that Zegnat asked...
#
GWG
"Do we actually have a start for that algorithm somewhere? A “which mf2 object to return for a given URL” algorithm?" - I asked where we should put such a thing
#
[tantek]
that's the mf2 parsing spec
#
KartikPrabhu
afaik, given a URL mf2 parsers returns the parsed mf2 of the entire page, not some "object"
#
[tantek]
interesting, I seem to have omitted it. Originally defined here: http://microformats.org/wiki/hcard-parsing#URL_handling
#
[tantek]
that's worth opening an issue for, especially as I think we already have implementations
#
KartikPrabhu
not sure any mf2 parser breaks up on fragments
#
KartikPrabhu
wait we have implementations for parsing only a fragment link?!
#
[tantek]
a URL with a fragment, yes
#
[tantek]
yes, feel free to open an issue https://github.com/microformats/microformats2-parsing/issues requesting incorporation of http://microformats.org/wiki/hcard-parsing#URL_handling into the microformats2 parsing specification
#
[tantek]
pretty sure both aaronpk and GWG use mf2 parser implementations that parse a specific element identified by fragment
#
KartikPrabhu
pin13 does not do it
#
[tantek]
could be a form UI bug
#
KartikPrabhu
must be something custom outside the actual parser
#
KartikPrabhu
I do that too
#
[tantek]
you need to test actual parser code, not just a wrapper UI
#
[snarfed]
interesting. i've had a consistent trickle of interest in bridgy for supporting fragments. problematic, but interesting. https://github.com/snarfed/bridgy/issues/445 , https://github.com/snarfed/bridgy/issues/681
#
Loqi
[csarven] #445 publish: support fragment for identifying individual entry
#
GWG
So, where is the algorithm discussed? And this is post parsing
#
[jgmac1106]
so I was trying to use a bit of javascript to get dt-updated, using this and getting close..but unreliable: https://answers.yahoo.com/question/index?qid=20091202150609AAC9bkV folks know of any other
#
[jgmac1106]
started searching Google when someone showed me: javascript:alert(document.lastModified) but that is just a popup modal so I started googling ways to call it HTML
#
GWG
I want to document the work Parse This does to standardize output and share it
#
GWG
Same for x-ray
#
Loqi
[Greg McVerry] Lesson One: Images, Headers, Links and Text
#
Loqi
[Greg McVerry] Lesson One: Images, Headers, Links and Text
#
[tantek]
GWG I linked to where it was described, and no it doesn't have to be after parsing, in fact you could do it much more efficiently if you go "pull out" the fragment element subtree first and just parse that
#
GWG
[tantek] I have a bunch of issues I would want to go over.
#
GWG
I was trying to find a place to document all of them for people writing similar code
#
[tantek]
GWG, see above re: my request to file an issue for that
#
GWG
Oh...
#
GWG
I thought you were referring to the fragment issue only
#
GWG
Excuse me
#
[tantek]
that's the only thing worth standardizing now right?
#
[tantek]
xray's output is an evolving thing. best bet for that is jf2
#
Loqi
[[tantek]] yes, feel free to open an issue https://github.com/microformats/microformats2-parsing/issues requesting incorporation of http://microformats.org/wiki/hcard-parsing#URL_handling into the microformats2 parsing specification
#
GWG
[tantek] I also have trying to standardize feeds in general
#
GWG
I did some work on the h-feed wiki page last night
#
GWG
I didn't realize someone added pagination as a recommendation
#
GWG
By the way, is it acceptable to update someone else's Indieweb example?
#
GWG
Such as what I did on the h-feed page?
phynite[m]1 joined the channel
#
[tantek]
yes and it's quite valuable - good work there GWG++
#
Loqi
GWG has 31 karma in this channel over the last year (158 in all channels)
#
aaronpk
XRay parses from a fragment url but I can't remember if that's XRay finding the fragment and handing just that part off to the mf2 parser or what