#dev 2018-02-17

2018-02-17 UTC
PeterFD and chrisaldrich joined the channel
#
www.boffosocko.com edited /presentation (+94) "reveal.js; moved TOC" (view diff)
#
KartikPrabhu ok so there does not seem to be any way to fork a github repo with its issues
#
www.boffosocko.com edited /follow (+85) "stub; emojicon suggestion 👣; microsub and Kinds of posts to see also" (view diff)
#
KartikPrabhu so I don't know how to import mf2py issues
renem joined the channel
#
www.boffosocko.com edited /posts (+22) "follow post kind to list" (view diff)
#
www.boffosocko.com created /👣 (+23) "redirect to following" (view diff)
#
aaronpk afaik you have to move it in order for the issues to come along
#
KartikPrabhu yeah
#
KartikPrabhu so I guess we'll have to get tommorries to move it
radicaledward101, tantek, snarfed, globbot, AngeloGladding, chrisaldrich, [mrkrndvs], [colinwalker] and [kevinmarks] joined the channel
#
Zegnat chrisaldrich: your use of `p-follow-of h-cite` is telling a mf2 parser that you want the value of the h-cite’s name property as the value for your follow-of property.
#
Zegnat I can see, especially for parsing it into a list of URLs, where you would want follow-of to be a URL. By using `u-follow-of h-cite` you would be telling parsers to take the url property of h-cite as follow-of value instead.
#
Zegnat Nested microformats can sometimes be a little tricky if you don’t know how they are getting parsed already :)
#
Zegnat chrisaldrich, is it easy for you to tweak that follow post layout?
#
Zegnat Oh, oof, you have one of those mf1 & mf2 mixing WP themes
jjuran_ joined the channel
#
Zegnat chrisaldrich, I would probably start of with something like this: http://php.microformats.io/?id=20180217114039677
sknebel and Zegnet joined the channel
#
www.boffosocko.com edited /follow (+646) "Chris Aldrich example" (view diff)
petermolnar_ joined the channel
#
Zegnat Huh, I wonder if this is still active
#
Zegnat https://rss-sync.github.io/Open-Reader-API/
#
sknebel doesn't look like it
#
sknebel (last github activity 2013, list link dead, ...)
petermolnar joined the channel
#
Zegnat Yeah, looks like the IRC channel is also dead
#
Zegnat [ChanServ] Registered : Jun 07 16:00:20 2013 (4y 36w 3d ago)
#
Zegnat [ChanServ] Last used : Nov 07 16:08:33 2014 (3y 14w 4d ago)
KartikPrabhu and leg joined the channel
#
GWG Zegnat, chrisaldrich is probably copying the way I set up some of my templates
#
GWG I may have a look now that I am making things more flexible.
#
GWG Right now, trying to design a duration picker
[kevinmarks] and pstuifzand joined the channel
#
Zegnat !tell aaronpk IndieAuth question. Can a server assume redirect_uri to have a scheme and path part available, or must it do some sort of normalization prior to adding to the query component?
#
Loqi Ok, I'll tell them that when I see them next
#
aaronpk good morning
#
Loqi aaronpk: Zegnat left you a message 16 minutes ago: IndieAuth question. Can a server assume redirect_uri to have a scheme and path part available, or must it do some sort of normalization prior to adding to the query component?
#
Loqi morning!
#
aaronpk hmm not sure how I missed mentioning that explicitly
#
aaronpk in the discovery section it says the endpoints (authorization and token) may be relative URLs
#
aaronpk https://indieauth.spec.indieweb.org/#discovery-by-clients
#
Zegnat Yeah, but not talking about discovery. Talking about redirect_uri as send by the client to the server.
#
aaronpk yeah, but it's all related anyway
#
aaronpk would be straightforward to resolve the redirect_uri (and client_id) from the URL they're fetched from
#
aaronpk the example I gave uses a relative URL in the HTML <link> https://indieauth.spec.indieweb.org/#redirect-url
#
aaronpk and it mentions "after resolving any relative URLs"
#
aaronpk so nevermind, it is explicit
#
Zegnat redirect_uri may be given as http://example.org, that does not have a path. So you would also want me to resolve non-relatives?
#
aaronpk yes because a URL without a path is not resolvable
#
aaronpk you have to add a path before you can do anything to it, like viewing it in a browser
#
Zegnat It just wasn’t clear to me whether my endpoint should “fix” the redirect_uri provided by the client or show an error
#
aaronpk ah
#
Zegnat I.e. clients are responsible for making sure the me parameter has a path (of at least /), so any me parameter sent by the client without a path throws an error.
#
Zegnat And I believe the same applies to client_id (MUST be http or https with host and path)
#
Zegnat But redirect_uri does not specify any such thing.
#
Zegnat Now that I think about it, not every scheme may require a path. Hmm...
#
Zegnat On that subject, aaronpk, I am guessing the redirect_uri send by the client in the auth request can not be relative right?
#
Zegnat That is, at minimum, it should consist of a scheme?
#
aaronpk it must be an absolute URL https://tools.ietf.org/html/rfc6749#section-3.1.2
#
Zegnat Aaah. Now I can start unraveling it :) Thanks
#
Zegnat absolute-URI = scheme ":" hier-part [ "?" query ]
tantek joined the channel
#
GWG Why do I seem to be spending the most time on a feature I was not using?
tantek and leg joined the channel
#
aaronpk hm, so miklb's site doesn't have an `author` property in the entries, although there is a very nice h-card at the top of the page
#
aaronpk https://miklb.com/
#
Loqi Michael Bishop
#
aaronpk according to the authorship algorithm, there is no link between that h-card and the h-entrys
#
aaronpk the "fix" is to either set an invisible author property on the h-entrys that matches the URL of the h-card, or wrap the whole thing in an h-feed so that the h-card can be the author property of the h-feed.
#
aaronpk ... or ... the authorship algorithm needs to be updated
#
GWG aaronpk, I prefer the h-feed option myself.
#
GWG I need to adjust mine again
#
GWG I miss the days when I asked Barnaby to look at it
#
Loqi misses the days when I asked Barnaby to look at it too
[jeremycherfas] and KartikPrabhu joined the channel
#
KartikPrabhu !tell snarfed: updated my fork of mf2py with latest code. Feel free to file issues https://github.com/kartikprabhu/mf2py/issues
#
Loqi Ok, I'll tell them that when I see them next
#
KartikPrabhu cc Zegnat
snarfed joined the channel
#
Zegnat I am not a fan of expending the authorship algo even further, aaronpk. Though if you find more sites that have this same problem, that is the neat solution.
#
aaronpk I know, it's already quite long as is
#
aaronpk it's a tradeoff between changing the consuming side or askign publishers to do something different tho
[kevinmarks] joined the channel
#
[kevinmarks] Part of the h-feed problem is that hAtom was missing feed level properties
#
Zegnat Which is why, if miklb is the only page we see this on, I think we should ask him to change it rather than chaning the algo.
#
Zegnat But if we find more pages where this is a problem, no other choice, I guess.
#
KartikPrabhu so much of mf2py has changed since I looked at it 4 years ago!
#
aaronpk oh yeah, adactio's links are another feed with the same issue
#
aaronpk https://adactio.com/links/
#
aaronpk a series of h-entrys with a single h-card at the same level, no author property on the entries
#
aaronpk his notes too https://adactio.com/notes/
#
aaronpk oh his home page too
[cleverdevil] joined the channel
#
sknebel one could try to use indiemap to check for this
#
aaronpk oh noes, together crashed again!
#
aaronpk https://reactjs.org/docs/error-decoder.html?invariant=130&args[]=undefined&args[]=
#
aaronpk grantcodes: do I just npm install?
#
grantcodes `yarn install`
#
aaronpk uhoh what is yarn
#
aaronpk https://yarnpkg.com/en/docs/install
#
grantcodes yeah that
#
[cleverdevil] I use homebrew for my node stuff. Makes life easier, IMO.
#
grantcodes I think it should really work with npm as well. Just not as guaranteed
#
snarfed KartikPrabhu++ for working on mf2py!
#
Loqi snarfed: KartikPrabhu left you a message 48 minutes ago: updated my fork of mf2py with latest code. Feel free to file issues https://github.com/kartikprabhu/mf2py/issues
#
Loqi kartikprabhu has 13 karma in this channel (165 overall)
#
grantcodes then `yarn run start`
#
aaronpk crosses fingers
#
snarfed (also oh noes now we have issue skew?! :( )
#
snarfed duplication--
#
Loqi duplication has -1 karma
#
KartikPrabhu wait what's issue skew?
#
grantcodes There were instructions on the readme I thought but I think [cleverdevil] removed them?
#
snarfed what is skew
#
Loqi It looks like we don't have a page for "skew" yet. Would you like to create it? (Or just say "skew is ____", a sentence describing the term)
#
snarfed skew happens when you duplicate a set of data, and often try to keep the copies in sync, but they never stay perfectly in sync, and inevitably they accumulate changes that conflict, and now you have n problems.
#
snarfed argh
#
KartikPrabhu snarfed: yeah there is no way to "copy" issues from a repo. So I manually added some with a URL linking to the original mf2py issue
#
snarfed skew is when you duplicate a set of data, and often try to keep the copies in sync, but they never stay perfectly in sync, and inevitably they accumulate changes that conflict, and now you have n problems.
#
Loqi ok
#
loqi.me created /skew (+232) "prompted by snarfed and dfn added by snarfed" (view diff)
#
snarfed KartikPrabhu: yeah i figured. why make copies at all though?
#
KartikPrabhu so I can see what to work on :P
#
snarfed ah ok. so, narrow, isolated, personal bookkeeping
#
KartikPrabhu yeah
#
snarfed you can't add labels to the existing issues since you're not a member?
#
KartikPrabhu yeah something like that. I don't have any access to tommorris' repo I think
#
snarfed anyway. your call. hopefully unnecessary once the repo is moved to the org!
#
aaronpk snarfed: hmm I am getting a weird result from granary, I tried using a new search term but it's showing me the results from a previous one I did
#
snarfed aaronpk: really!
#
aaronpk actually weirder, the form preview shows an empty <body></body> but when I view it in the browser I get the other results
#
snarfed huh!
#
snarfed it does cache many silo results for 5m, but that leaking obviously shouldn't happen. feel free to file a bug if it's reproducible
#
aaronpk it seems to be happening consistently with this one search term
#
snarfed the web UI needs a progress spinner for when a request is outstanding, that's confusing right now
#
snarfed but not others?!
#
aaronpk actually no i'm super confused
#
aaronpk when I view the URL in a browser, i'm getting the same search results no matter what I put as the search term
#
aaronpk if I use the HTML form then I get correct results with the exception of this one search term which is empty
chrisaldrich joined the channel
#
chrisaldrich Zegnat++ for catching my error. Thanks for the help!
#
Loqi zegnat has 46 karma in this channel (172 overall)
#
snarfed aaronpk: ah yes, iirc there may be some caching that only the web UI uses
#
snarfed i've forgotten my logic for that, seems silly and obviously confusing now :P
#
snarfed actually no i think it was the opposite
#
snarfed ugh, checking
#
snarfed aaronpk: try adding &cache=false to the URLs
#
aaronpk that worked
#
aaronpk the one search term is blank still tho. "streampdx"
#
snarfed it may have cached a timeout
#
snarfed oh even with cache=false?
#
snarfed looking
#
snarfed i see the requests. logs seem to show they're doing the normal thing, using the API. let's see...
#
grantcodes Who should I message about the JF2 spec doc? Think I noticed an error.
#
aaronpk there should be an issues link
#
aaronpk https://github.com/indieweb/jf2/issues
#
[cleverdevil] Yeah, the instructions weren't working anymore, so I removed them ;)_
#
[cleverdevil] I'll try and add some working instructions back later today.
#
aaronpk is there some config thing I need? tried `yarn run start` and got this: SyntaxError: Unexpected token )
#
snarfed aaronpk: sigh, as expected, twitter search API is returning zero results for streampdx
#
snarfed https://brid.gy/about#Twitter+responses
#
grantcodes Ahh yeah issues link. Thanks.
#
snarfed oh actually because the (not new) twitter search API is only last 7d
#
grantcodes Don't think you should need to do anything apart from install and run
#
snarfed ok. i still need granary's caching since the feeds it serves are high traffic, but i'll rethink it to be less confusing. and i'll look into the basic tier of twitter's new search API that supports full history.
#
aaronpk ah bummer
#
aaronpk well seems like the caching should at least take the search term into account
#
KartikPrabhu snarfed: you want the implied p-name thing yes?
#
snarfed KartikPrabhu: yes! i don't fully understand what it is, but i want it
#
KartikPrabhu ha! ok
#
snarfed aaronpk: yeah the cache leak is clearly bad, and i definitely don't understand it yet
#
snarfed (cache key is full URL)
#
snarfed aaronpk: to confirm, you tried one search term, then another, and got the same results?
#
KartikPrabhu 's brain has forgotten the mf2py code flow
#
snarfed KartikPrabhu: basically i suspect better implied p-name handling is the right fix for https://github.com/snarfed/granary/issues/131
#
Loqi [aaronpk] #131 Instagram->HTML: photos with no caption are missing a blank name property
#
[cleverdevil] FWIW, I also get an error on a clean install now, [grantcodes].
#
KartikPrabhu snarfed: ok. that discussed fix is tricky as it is sort of a "global" check
#
[cleverdevil] ```Error running install script for optional dependency: "/Users/cleverdevil/Development/together/node_modules/chokidar/node_modules/fsevents: Command failed.```
tantek joined the channel
#
grantcodes Ahh that's just when running `yarn install`?
#
[cleverdevil] Yup, that said it looks like things still work in spite of that.
#
snarfed ugh, i have to ask for permission for the new twitter search API, and describe what i'm building
#
snarfed https://developer.twitter.com/en/apply/usecase
#
snarfed knows how this goes
#
tantek what is yarn?
#
Loqi It looks like we don't have a page for "yarn" yet. Would you like to create it? (Or just say "yarn is ____", a sentence describing the term)
#
snarfed can't tell if this is just because it's in beta right now, or permanent
#
[cleverdevil] yarn is a package and dependency management tool for node.js - https://yarnpkg.com/
#
Loqi ok
#
loqi.me created /yarn (+111) "prompted by tantek and dfn added by [cleverdevil]" (view diff)
#
tantek re: authorship, I think I agree with what folks were saying - better to ask [miklb] about how much work it is to add an h-feed around the page that incorporates the existing h-card in a p-author
#
tantek at the same time, start collecting examples of the "page of h-entrys without author, but with a representative h-card"
#
[cleverdevil] (Instructions added back to the README for Together, btw.)
renem joined the channel
#
tantek.com edited /Planning (-430) "/* Los Angeles */ remove past dates" (view diff)
#
[cleverdevil] Added links to avatars and author names for ya [aaron_pk].
#
KartikPrabhu in support of databases :https://allinthehead.com/retro/362/why-use-a-database-with-a-small-cms
#
Loqi [Drew McLellan] Why Use a Database With a Small CMS?
#
KartikPrabhu post also uses Webmention so...
#
[cleverdevil] [grantcodes] [aaron_pk] what do you think of this look for like-of? https://cleverdevil.io/s/rmNqmtNyW6.png
#
[cleverdevil] (I've done similar for bookmark-of, in-reply-to, repost-of, etc.)
gRegorLove joined the channel
#
tantek from drew's post: "or some rickety direct-editing system working directly on the files" lol. ahem. :)
#
sknebel vim isn't "rickety" :P
#
sknebel (the human behind it trying to get the custom format right apparently is occasionally though...)
#
tantek I think that's the point
#
tantek about once a month I break my site because of an editing error in the storage files, usually fixed in under a minute.
#
tantek pretty trivial compared to all the upsides so I tolerate it
#
KartikPrabhu I don't think I have ever broken my site due to storage error. Code errors yes
#
KartikPrabhu also that post by Drew does support database as cache
#
tantek indeed
#
tantek also if you're designing some PHP for *other people* setup on a *random webhost* - the fact that webhosts already have UIs for managing MySQL means you can offload the burden onto those other people (e.g. backups etc.)
[jeremycherfas] and barpthewire joined the channel
#
[cleverdevil] Getting pretty close to adding a bookmark-of action to Together.
#
@ellemthomas24 my life right here at the moment #amediting #amwritingromance #indieauth #selfpublishing https://twitter.com/byMorganWright/status/964935600594878466 (twitter.com/_/status/964979005437939713)
snarfed, Kongaloosh and AngeloGladding joined the channel
#
[cleverdevil] Whew. Well, I got bookmark-of posts pretty much working in Together, though there is a weird error after the post is created on the Micropub endpoint.
#
[cleverdevil] !tell grantcodes take a look when you get a chance - https://github.com/cleverdevil/together/issues/16
#
Loqi Ok, I'll tell them that when I see them next
#
Loqi [cleverdevil] #16 Ability to repost with comment
tantek joined the channel
#
KartikPrabhu !tell snarfed: pushed the implied name fix https://github.com/kartikprabhu/mf2py/tree/implied-name-fix let me know if it works for what you want
#
Loqi Ok, I'll tell them that when I see them next
[tantek] joined the channel
#
[tantek] Good approach Kartikprabhu, seeing is the updated parser solves the problem when integrated into a production system
#
KartikPrabhu yup
#
[tantek] Going to try once again to get posse to Github working this weekend
#
[tantek] Going to look forward to testing things like images
[miklb] joined the channel
#
[miklb] tantek ask me about what again?
KartikPrabhu and [kevinmarks] joined the channel
#
[kevinmarks] I wonder if implied name fix will break post types
#
[kevinmarks] Because name == content is a test there
#
[tantek] Now I can’t remember
#
[miklb] cool. I saw my nick highlighted in Slack scroll back but didn’t see context. Something about h-feed
#
[tantek] Kevinmarks no because that test is there from notes with explicit p-name and e-content on the same element
#
[miklb] Still haven’t quite figured out notifications for IRC->Slack nicks
#
[kevinmarks] https://indieweb.org/post-type-discovery#note_detection_name_content_details
#
[tantek] Kevinmarks, explicit not implied was the use case there
#
[tantek] So no, I don’t expect any breakage from changing implied p-name
#
[tantek] Especially since in real cases with e-content the annoying thing is that implied p-name would be *more* than the content
#
[tantek] Hence why it made sense to have e-* prop(s) also stop implied p-name
#
[kevinmarks] Right, and rereading it, no name or empty name is already handled
#
[tantek] [miklb] I think it was about marking up your home page with explicit h-feed around your h-card and h-entry s
#
[tantek] WDYT?
#
[miklb] 🎉 Slack notifies me of notifications of [miklb] from IRC in Slack if I’m not in the app
#
[miklb] What is WDYT
#
Loqi It looks like we don't have a page for "WDYT" yet. Would you like to create it? (Or just say "WDYT is ____", a sentence describing the term)
snarfed, tantek and [cleverdevil] joined the channel
#
[cleverdevil] Wooo! Added a composer to Together for status updates.
#
[cleverdevil] https://cleverdevil.io/s/dTD1g7DnJV.mp4
#
KartikPrabhu [cleverdevil]++ looks neat
#
Loqi cleverdevil has 9 karma in this channel (59 overall)
#
aaronpk wow nice
#
aaronpk but.. what is repost with comment? I thought repost is just a repost. If you add a comment it's not a repost, it's a post with a quotation
#
aaronpk It appears I have already said that in the github issue :-)
#
snarfed points for consistency
#
Loqi snarfed: KartikPrabhu left you a message 1 hour, 31 minutes ago: pushed the implied name fix https://github.com/kartikprabhu/mf2py/tree/implied-name-fix let me know if it works for what you want
#
[cleverdevil] Its not a repost-with-comment.
#
[cleverdevil] Its a bookmark-of 🙂
#
[cleverdevil] That you can give a title and content to.
#
[cleverdevil] (But, that's a separate issue).