#dev 2017-11-24

2017-11-24 UTC
#
tantek looks like Googlebot is now my top bandwidth requester
#
GWG tantek: Which it? The original user agent talk, or my comment that the URL in question couldn't be interpreted.
#
GWG aaronpk: But it has an h-feed and a non-nested h-entry. That is a case neither you nor I envisioned.
#
tantek GWG, technically it has hfeed and hentry (mf1)
#
GWG tantek: Yes. I know. But should a parser account for improper feeds.
#
tantek GWG, perhaps a parser client should
#
GWG That is what I'm looking at.
#
tantek but an mf2 parser no. it should just follow the mf2 parsing spec
#
tantek in this case, the h-feed is empty, likely due to a nesting error
#
GWG tantek: Mine is a client in that scenario.
#
tantek so that's an easy case to detect and either ignore (easiest), or assume it's an h-feed for the rest of the h-entrys on that page
#
GWG PHP-MF2 is my parser.
#
tantek in this case the hfeed adds no information, so it can be safely ignored
#
GWG I'm trying to figure out how to write the rule.
#
GWG According to the logic I have, it is being rejected because it is checking for a matching url property.
#
tantek if the h-feed has no .children, or .children.length==0, then ignore it
#
tantek as if you never saw it
#
GWG tantek: My logic is. But if there are multiple items, it checks each item for the presence of a URL property.
#
GWG The hfeed has one, but the hentry does not.
#
GWG That is what I'm trying to figure out how to account for without messing up the proper logic.
#
tantek GWG having a URL is not good enough. if the h-feed has no children you can ignore it before you even check for presence of any URL properties
John___ joined the channel
#
GWG Okay, Blogspot is also doing the non nested hfeed thing.
snarfed joined the channel
#
snarfed !tell tantek thanks! i actually fixed it the day after it happened. https://github.com/kylewm/brevity/issues/8 . tldr: when they launched 280 chars, twitter also substantially changed the way they *count* chars. https://twittercommunity.com/t/updating-the-character-limit-and-the-twitter-text-library/96425/1
#
Loqi Ok, I'll tell them that when I see them next
#
Loqi snarfed: tantek left you a message 2 hours, 23 minutes ago: ICYMI - looks like I found a bug with Bridgy Publish's support of longer tweets when there's a photo included: https://chat.indieweb.org/dev/2017-11-18/1510993441166000 I couldn't find an issue on this in particular, should I file a new one?
#
Loqi [snarfed] #8 Recent Twitter char counting changes ?
renem and tantek joined the channel
#
tantek snarfed wow just wow
#
Loqi tantek: snarfed left you a message 48 minutes ago: thanks! i actually fixed it the day after it happened. https://github.com/kylewm/brevity/issues/8 . tldr: when they launched 280 chars, twitter also substantially changed the way they *count* chars. https://twittercommunity.com/t/updating-the-character-limit-and-the-twitter-text-library/96425/1
tantek, jjuran and tbbrown joined the channel
#
GWG Okay. Parsing issues fixed.
tantek, [xavierroy], cweiske, [mrkrndvs], loicm and marcthiele joined the channel
#
cweiske.de edited /comments (+123) "/* Comments Embeds */ github-comments" (view diff)
marcthiele, marc_thiele, tantek, John___, davidmead, [khurt], snarfed and jjuran joined the channel
#
@fhsales Check out today's new blog post! Webmentions: Pingbacks For The Modern Web http://bit.ly/2mZjHNk (twitter.com/_/status/934097017596465152)
John___ joined the channel
#
GWG Parts of that article feel rather negative
#
sknebel seems that they don't believe that off-silo comments are that important, so they are kinda negative on that aspect. or did you mean something else?
jjuran joined the channel
#
GWG They seem negative on the concept in general
#
Zegnat I think you may get that feel because it feels a bit like a fluff piece with carefully placed SEO links.
#
Zegnat E.g. they say “the IndieWeb movement feels something important was lost with the decline of comments and conversations between bloggers and publishers, each with their own domain.” But they never point out what the important thing that was lost is: control of the content, less content loss when websites go down.
#
Zegnat Probably never were interested in going into anything.
#
aaronpk also the hyperlinks to their hosting solutions
#
aaronpk doesn't seem super negative to me overall tho
marcthiele, snarfed and davidmead joined the channel
#
Zegnat Yep, not negative, just ... shallow? It isn’t really an article that explains any of the tech. It is just a post. I guess a way to keep your corporate blog going?
davidmead joined the channel
#
Zegnat Interesting that an escort firm can write a better blog post about Webmentions than a hosting company ;)
#
aaronpk grantcodes: does this workflow for paging through a timeline make sense to you? https://indieweb.org/Microsub-spec#Example_Paging_Workflow
#
GWG Zegnat: Leads to some great Indieweb pickup lines, doesn't it?
davidmead joined the channel
#
aaronparecki.com edited /Microsub-spec (+191) "/* Example Paging Workflow */" (view diff)
[miklb], John___ and KartikPrabhu joined the channel
#
aaronparecki.com edited /Microsub-spec (+497) "/* Paging */" (view diff)
#
aaronpk tests++
#
Loqi tests has 4 karma in this channel (13 overall)
#
aaronpk GWG: just noticed your checkins have the foursquare URL as your u-url property. it'd be better if the foursquare URL was a u-syndication class
#
aaronpk oh wait actually no it's worse, you're setting the venue URL as the u-url of the h-entry
#
GWG aaronpk: I'll have a look. The Swarm URL is definitely a u-syndication.
#
GWG But, it must be how Post Kinds is parsing the Micropub data
#
aaronpk yep I see that now
#
GWG I just released the first check-in code to the public
#
GWG I'm already working on the bugfixes from version 2.7.0, which was a bunch of massive structural changes
#
GWG So, I'll see about why it is off.
#
GWG It is a template, so it will propagate back to older posts
#
GWG aaronpk: In order to get anything to display, I had to change how the whole system stored mf2 properties
#
aaronpk crazy
#
GWG I had to switch from jf2 to mf2 json
marcthiele joined the channel
#
grantcodes aaronpk: Hmm think the paging looks good. In your examples it looks like you have 2 values for both `after` and `before` separated by a semi colon or do your post uids have a semi colon?
#
grantcodes Also I assume the number of 10 per page is changable? Seems quite low for a default to me
tantek joined the channel
#
aaronpk grantcodes: the cursor values are opaque to the client so it's not important what my examples represent ;-)
#
aaronpk yeah 10 is probably low but I also dont' think that needs to be defined by the spec
#
grantcodes Yeah don't know if this is meant to be a techical but if it said `&after=post100&before90` or something more readable it would make more sense to me
#
grantcodes Might be a good idea to have a set default just so there is more consistency accross readers?
#
tantek joining late, but did someone collect examples of such query params?
#
tantek e.g. Mediawiki searches
#
aaronpk I could update the example to use sequential numeric IDs, but it's also meant to be opaque to clients so that clients don't make assumptions about the values
#
aaronpk https://indieweb.org/pagination#APIs
#
tantek https://indieweb.org/wiki/index.php?title=Special:Contributions&offset=20171028210936&limit=100&target=Tantek.com
#
tantek.com edited /pagination (+257) "/* APIs */ add MediaWiki" (view diff)
#
aaronpk hm interesting, that's not actually part of their API
#
tantek I view query params as part of the API of a site
#
aaronpk i'm specifically talking about API design, so I disagree
#
tantek I believe the query params were designed
#
aaronpk this is a useful example for website paging, but it's not part of their API
#
tantek it's over HTTP, so I'm not sure of the distinction you're making
#
aaronpk considering they have a separate API with a different paging mechanism, I want to capture that as well https://www.mediawiki.org/wiki/API:RecentChanges#Parameters
#
tantek they're both APIs
#
tantek just different content types being returned
#
tantek Google uses "start", e.g. https://www.google.com/search?q=indieweb&start=100
#
tantek.com edited /pagination (+174) "/* APIs */ Google Web Search / start = n" (view diff)
#
tantek "separate API", sigh.
#
aaronpk for retrieving the source wiki markup, and making edits
#
tantek classic web devs can't figure out how to use HTML as an API so they offer a smorgasbord of other rando formats instead and call it an "API" as if changing the syntax (format) changes the semantic (what's an API or not) https://www.mediawiki.org/wiki/Special:MyLanguage/API:Data_formats#Output
#
aaronpk HTML isn't an API for wikis since you don't write HTML when you edit wiki pages
#
aaronpk so you're always going to need some way to get the source data behind a mediwiki page
#
tantek aaronpk - when you *edit* a wiki page it sends you back the wiki markup for example
#
tantek there's also a "view source" option to query a mediawiki page and get back the wiki markup
#
tantek all in HTML
#
tantek these other rando formats are only because webdevs couldn't figure out how to re-use the existing HTML "API"s
#
aaronpk too bad that's really hard to write code against
#
tantek only if the HTML produced is crap, which would be the same problem if the XML produced is crap, or the JSON produced is crap, or the YAML etc
#
tantek producing well formed HTML, even XML-valid HTML, is no harder than producing valid XML
#
aaronpk producing well formed JSON is way easier tho
#
tantek and yet they go to all this duplicate effor to do so (a whole another format), instead of doing the significantly *less* incremental work of producing higher quality HTML
#
aaronpk anyway this is not really a useful debate, this ship sailed a long time ago
#
tantek MediaWiki in particular never had much of a semantic HTML cuture
#
tantek *culture
#
tantek note the long list of rando formats they went through and abandoned (link) above
#
tantek wddx, dump, txt, dbg, yaml
#
tantek ^^^ that's the ship that sailed a long time ago
#
tantek and evidence of why that "ship a different format" methodology is fundamentally flawed (more work to maintain)
#
tantek how many more extra formats do they have to build then abandon before they learn?
#
aaronparecki.com edited /pagination (+417) "/* MediaWiki */ include info about the MediaWiki API" (view diff)
#
tantek xml is deprecated, and maybe their JSON output will be deprecated in another 10 years for whatever new dev-centric format is fashionable in the next 5 years
#
tantek but they'll still be supporting HTML
#
aaronpk anyway this conversation is not about HTML vs other formats for APIs
#
aaronpk paging has nothing to do with the serialization format returned
#
tantek right, which is why my initial claim stands - the search queries are API params
#
www.svenknebel.de edited /pagination (+112) "/* MediaWiki */ API" (view diff)
#
tantek whether they return HTML or something else
#
aaronpk nice find sknebel
#
aaronparecki.com edited /pagination (+1205) "/* APIs */ Twitter" (view diff)
#
tantek I do like the queries that use datetimes for offsets instead of ordinal numbers
#
aaronpk definitely
#
tantek seems like people are slowly learning that lesson - I see more and more of them
#
tantek except in cases where the service has nearly no datetime knowledge, or crappy knowledge thereof, like Google Searches
#
aaronpk grantcodes: i'm hesitant to recommend any specific default page size cause I feel like it might be dependent on the context. but maybe I could use 20 in the examples just to sort of set the baseline?
#
aaronparecki.com edited /pagination (+269) "/* Twitter */" (view diff)
#
aaronpk this "cursor" pagination concept that's appearing in a bunch of places is essentially an abstracted solution for date-based pagination
#
aaronpk if the things you're paging over have dates, then often you'll see the cursor values are actually dates.
#
tantek kinda. it's an abstraction to be able to incorporate or use other stable indices besides dates
#
aaronpk yeah
#
tantek it's also well established in database API cultures of 25+ years and more. pre-web
#
tantek but if you've got dates, might as well expose them
#
grantcodes aaronpk: yep I think 20 is a better example number. And for the cursors I'm no expert but is it not more useful to use uids just in case your data is sorted by something other than date? (not a hugely likely scenario). In the example I would remove the semicolons just so it's more obvious it's a random id
#
aaronpk okay
#
aaronpk those are actually the strings I use in Monocle, it's actually two timestamps that I use to sort. date added to the channel and also published date.
#
aaronpk but the nice thing is it's implementation specific so you don't need to know that :)
#
aaronparecki.com edited /Microsub-spec (-9) "/* Example Paging Workflow */ update example to 20 per page, and remove colon so cursors are just a hex string" (view diff)
#
aaronpk another question, should servers be expected to support a "limit" or "count" query param so the client can decide how many results it wants per request?
marcthiele joined the channel
#
tantek hex string? why not NewBase60 string? (they are URL safe by design :D )
#
aaronpk heh, dechex() was just sitting right there
#
aaronpk let's see what it looks like with newbase60
#
tantek I happen to like chex, why is there a function to remove it?
#
aaronpk nice and short: 1wcr11:1wc_L1
#
grantcodes I would say a count/limit param would be useful, but it would probably need to be limited. I know the WordPress API has a max of 100 for example
#
aaronpk yeah i'd expect each server would set its own upper limit
#
grantcodes In that case it would need to handle requests for more than its upper limit but return less posts than were requested
#
aaronpk yeah, it'd be able to tell if there was more data from the paging keys
#
aaronpk so if you request 1000 items but the server maxes out at 100 at a time, it would return 100 items and a "after" cursor to get the next 100
#
grantcodes Yeah that makes sense. The only issue is I can totally imagine people writing a for 1 to 1000 loop and thinking it works fine on with their microsub server but breaking with other servers that won't return 1000 posts. But I might be trying to idiotproof it too much ?
#
[miklb] do not underestimate the ingenuity of idiots
#
aaronpk heh yeah, if you're doing for i = 1 to 1000 instead of a foreach then I don't really want to get in your way of shooting yourself in the foot
#
grantcodes Maybe in the spec it would be good to say the server must support a count from 1 to 100 (as an example) but may support more if they want
#
@mrfabiobarbosa Testing #webmentions on my personal website: https://fabiobarbosa.net/blog/letter-young-mystic (twitter.com/_/status/934152051411619840)
#
aaronpk i'm kind of inclined to specify as little as possible about things that don't actually affect interop
#
aaronpk so if a server wants to return only one item at a time, as long as it supports the paging parameters a client will be able to deal with it without any different code than a server that supports 100 at a time
#
grantcodes The thing I dont like about that is different experiences for users in the same client.
#
grantcodes If it supports a count you know all the users of a client will see x posts at a time
#
aaronpk I was imagining clients working more like slack clients where it just keeps loading history of a channel seamlessly just might do it different numbers of chunks at a time
#
aaronpk The client doesn't have to actually present paged data
#
aaronpk I think it'd be reasonable to recommend a server supports page sizes of like 20-100 but clients should be able to handle whatever comes back and ultimately can't make assumptions about the server anyway
#
grantcodes yeah the client make assumptions anyway if it is at the end of the dataset
#
grantcodes *can't make assumptions
loicm, Loqi, cweiske and jjuran joined the channel
#
aaronparecki.com edited /Microsub-spec (+619) "/* Paging */ limit" (view diff)
#
tantek.com edited /next-hwc (+0) "next 11-29" (view diff)
[miklb] joined the channel
#
KartikPrabhu aaronpk: why "before" and "after" in the micro-sub pagination with explanation to "newer" and "older"? Why not use "newer" and "older" directly as property names?
#
aaronpk "newer" and "older" only make sense for pagination of things that have timestamps, whereas "before/after" can refer to positions in a list, not implying timestamp ordering
#
KartikPrabhu aah, then the explanation might need tweaking
#
aaronpk hm I should update that description
#
aaronpk I started writing it in the context of the "timeline" example but that's not obvious now
#
KartikPrabhu yeah in timeline "newer" and "older" seem more self-explnatory
#
aaronparecki.com edited /Microsub-spec (-4) "/* Paging */ remove "newer/older" phrasing from paging description" (view diff)
#
tantek.com edited /Category:syndication (+133) "expand dfn for what is, no need for empty see also" (view diff)
#
tantek.com moved /Cross-posting to /cross-posting "lowercase non proper noun"
#
tantek.com edited /cross-posting (+25) "/* See Also */ syndication-models" (view diff)
KartikPrabhu and marcthiele joined the channel
#
tantek.com edited /Main_Page (-227) "/* Homebrew Website Club */ update for next week" (view diff)
#
aaronparecki.com edited /Microsub-spec (+959) "/* Following */ following and unfollowing" (view diff)
#
aaronparecki.com edited /Microsub-spec (+93) "/* Search */ return "type" in search h-feed items" (view diff)
#
aaronparecki.com edited /Microsub-spec (+162) "/* Following */ return jf2 h-feed objects" (view diff)
snarfed joined the channel
#
aaronparecki.com edited /Microsub-spec (+94) "/* Preview */" (view diff)
KartikPrabhu joined the channel
#
sknebel what is logo?
#
Loqi These are the official versions of the IndieWebCamp logo https://indieweb.org/logo
#
aaronpk !tell grantcodes I just finished up enough of Monocle that it should be a functional prototype! I've added you to the beta so you can log in to it! https://monocle.p3k.io/dashboard
#
Loqi Ok, I'll tell them that when I see them next
#
snarfed aaronpk++
#
Loqi aaronpk has 90 karma in this channel (1477 overall)
marcthiele joined the channel
#
grantcodes aaronpk: awesome! I'll start playing with it next week
#
Loqi grantcodes: aaronpk left you a message 11 minutes ago: I just finished up enough of Monocle that it should be a functional prototype! I've added you to the beta so you can log in to it! https://monocle.p3k.io/dashboard