grantcodesI don't understand how an entry can't have a url? I can see it not having a unique url, but it must have been retreived from a url somewhere? But apart from that I don't know a solution.
aaronpkgrantcodes: right now, i'm not actually fetching any post permalinks, i'm only parsing entries from an h-feed page, so it's entirely possible there's an h-entry there with no URL
grantcodesWhat I can see being very useful in the long term is for a way for clients to add metadata to posts on the microsub server. You can say send a update request to mark a entry as the last read entry or it has been liked or replied to.
aaronpkI think that's part of the legacy of treating RSS readers as an email-like interface, with the display of number of unread items which everyone grew to dislike
grantcodesRE: Image resizing, I think it might be ok as an optional feature to a microsub server. I love when apis or whatever have on demand image resizing so the client can request whatever width & height they want. It can also be easily done with graceful degredation, if a client requests micropubserver.com/image.jpg?height=100&width=100 the server could read the query and resize or it could do nothing and return the full
aaronpkmost client-side environments can handle arbitrary size images though, right? it's just a matter of reducing the download time really. you can stick a 3000x6000 pixel image into an <img width="100"> tag and the browser will resize it
grantcodesaaronpk: not necessarily per item but I think the last read item is pretty vital. Noone likes scrolling back through a feed to find their last read
aaronpkthere's always the question of do you crop or fit, how do you indicate which direction is okay to crop, etc. just look at the syntax for imagemagick's resize functions, it's super ugly
grantcodesIt has options for width and height. And the cool one is letterboxing, which is where it adds extra whitespace to fit the entire image inside a specific aspect ratio if that is what you want
aaronpkit does seem like image sizing is something the microsub server could help out with, especially since it also means the server could cache/store images so they're available for readers even if the originals are gone
grantcodesYeah my use case is I host full size images at around 8mb and galleries of 100+ images. You do not want to download those full size on your mobile internet ?
aaronpktantek: yes, but the next step is that the client needs to know whether or not a post has already been liked on your website, and it's unreasonable to make the client fetch your website to check, since there isn't a good way to even do that
aaronpkthe microsub server could be subscribed to your website to find out whether you've liked a post, but now we need some way to indicate that to the client
grantcodesWith the microquery (or whatever) the problem then is you now have an extra set of requirements, you need a microsub server and your site must support this not trivially easy to set up microquery system in order to have a good user experience
aaronpkyeah I think the simple version is the microsub server can subscribe to your site and track all your likes/replies/etc that appear there, then include that information in the items it returns to the client
[cleverdevil]That’s a lot of noise when it could be elegantly handled by the Microsub server and wouldn’t require someone to have a Micropub compatible website to use a reader.
tantek[cleverdevil], it's a simplifying assumption. you have to do the work somewhere to store/retrieve "stuff", that *is* a micropub server, which makes the most sense to be your own server
grantcodesBut when you do like something the microsub server would still have to store it as you might move clients or you would have to scrape your website for those likes every time you open a client
aaronpkgrantcodes: yeah the microsub server needs to be made aware of the "like" that was created via a micropub request. how that actually happens is an implementation decision, since someone's micropub and microsub server might be the same.
grantcodesWell in terms of MVP my personal priorities of what is being discussed would be 1. Feed subscription & retrieval (of course) 2. Read state / maybe more channel state variables 3. /response saving 4. Image resizing
tantekif a bunch of indieweb sites use the same microsub server, and happen to subscribe/follow the same feeds, the microsub server gets to request those feeds *once* instead of once per client
grantcodesaaronpk: Only other thing that hasn't been discussed that I can think of for storing in state would be something like the tweetdeck filters for channels. Eg. I want a channel that is only actual content from people I follow. Not likes, reposts, bookmarks etc,
tantekI think this is more of a UI problem, and that by solving the UI problem (with multiple examples), you might be able to find commonalities that a protocol feature would make more efficient
grantcodesAgain has that same benefit that you can then change your filters but all the posts that were hidden are still stored on the microsub server and can be shown
tantekI guess because I'm more interested in the "your website is your reader" use-case, I'm less interested (lower, not zero, priority) in questions of "preserved between clients"
aaronpkone of the UI goals I have is to have completely different layouts for presenting posts, where some of the layouts only make sense for certain post types, e.g. a photo grid wouldn't even show posts without photos
aaronpki'm going to make monocle return cached data for the preview if it's already been fetching that feed, and if it hasn't seen it yet it'll actually go and fetch the feed to preview it
grantcodesHow would you expect to handle the channels thing? Every subscription is added to the home channel correct? Then you can optionally add it to other channels
aaronpkgrantcodes: that's the idea with "default", if you don't select one then it just goes to the default channel. i'm expecting the vast majority of people to just have that one channel like a twitter home timeline
aaronpktantek: what I liked about this UI https://indieweb.org/follow#Feedly is you can just type "tantek.com" and click "follow" on the first result. nowhere does it even mention "feed" in the UI
aaronpk(okay there are two places it mentions feeds but they are inconsequential. one is the little grey "feed" on the preview, and the other is in the "follow" dropdown menu but that is confusingly a different meaning of "feed")
www.boffosocko.comedited /h-card (+831) "h-card generator; link to additional examples; expand definition; category: microformats; How to on WordPress" (view diff)
grantcodesOnly problem I see with it is it might be hard for servers to return a decent list of results without the huge existing dataset that I'm sure feedly already stores.
tantekone possible hack for those is, if they're not HTML, if they have a rel=alternate back to an HTML page which then has a rel=alternate back to them, then they're public
tantek(though it's possible they may rel=alternate to an HTML capability URL, more likely they'll rel=alternate to the "plain" HTML page, which if it works without auth to provide a rel=alternate link back to the feed, then the feed is likely public)
grantcodesI'm sure you will, but if you separate that "search" functionality out so it would be extendable that would be awesome. Can see it continually improving, like searching by Twitter handle for example
Loqiautosuggest (AKA auto-suggest) is a user-interface feature that provides a list of options while the user is typing, related to what the user has typed, possibly beyond mere (sub)string matches https://indieweb.org/autosuggest
ZegnatNot only did I not remember the use-case for empty href, I also don’t remember why I didn’t just PR the fix myself aaronpk. Sorry for that one, haha
ancardaI don't think it will; CAA records are checked by CAs when they go to issue a certificate - it's suppose to help with mis-issuance and make it harder for an attacker to get a certificate by limiting them to CAs you trust, but security products will make one on the fly
ancardaZegnat: If you check Namecheap, they might allow CAA records to be created. You'd need one called "issue letsencrypt.org" (make a new `issue` record per CA you want to whitelist). You can also add "iodef mailto:your@email.address" which will email you when there's an attempted certificate request that was blocked
ancardaCAA is an acronym for Certification Authority Authorization, a DNS record type that indicates what Certificate Authorities are allowed to issue certificates for a domain. It supports whitelisting certificates, wildcards, and can also send reports of attempted requests that were blocked.
Loqiok, I added "https://gist.github.com/roycewilliams/1710ade469c05eb0b090d268470aa741 (List of supported DNS software and hosting providers)" to the "See Also" section of /CAA
ancardaKartikPrabhu: Yeah, you're right. I'm just new to the community and I don't want to step on anyone's toes. Although, the line "This article is a stub. You can help the IndieWeb wiki by expanding it." basically says the wiki welcomes edits
loqi.meedited /CAA (+87) "ancarda added "https://sslmate.com/caa/ (CAA record generator for a variety of DNS server software)" to "See Also"" (view diff)
Loqiok, I added "https://tools.ietf.org/html/rfc6844 (RFC6844 - DNS Certification Authority Authorization (CAA) Resource Record)" to the "See Also" section of /CAA
LoqiTantek-ing is a method of encouraging people to contribute to the wiki by indirectly prompting the person who first mentioned the term to create a short wiki dfn page for it https://indieweb.org/tanteking
ZegnatDon’t think there is a way around that, aaronpk? Guess you just have to have WebSub on your site so you don’t need to do polling but can wait for the feed to tell you about significant changes.
Loqisecurity in the context of the indieweb may refer to security concerns regarding personal domains, web hosting, https setup, private data, identity etc https://indieweb.org/security
ancardaThanks, hopefully this page is ok? https://indieweb.org/CAA -- could someone tell me if it looks alright and if you want more detail or anything?
GWGThe new Press This plugin in WordPress is really helping me refine my own parsing code...my xray equivalent. I test every URL someone there complains about not working.
aaronpkmy nginx config basically has upstream php5 { server unix:/var/run/php/php5.6-fpm.sock; } and php7 { server unix:/var/run/php/php7.1-fpm.sock; }
GWGI have to ask all of you. If a site you are trying to respond to returns a non 200 because it is filtering your retrieval of it, let's say by user agent...what should your site do? This applies to webmentions, and to my case...the link preview I'm trying to generate.
aaronpkmy reasoning is these things happen because people run plugins intended to stop certain types of bots, and my use of that site does not fall under the type of thing they're trying to block, so I work around it
tantek!tell snarfed 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?
tantekEvery browser puts "Mozilla" in their UA string since Netscape started to and sites started only sending "rich content" to Netscape via checking for Mozilla.
[kevinmarks] Er 304 dyac
Loqi etag is a header sent in an HTTP response to allow future fetches of the same URL to be skipped if nothing has changed https://indieweb.org/etag
www.boffosocko.com edited /Indieweb_for_Journalism (+1692) "Additional material on syndication" (view diff)
grantcodes Yes! Loving the microsub chat!
aaronpk http://websynthesis.com/timthumb-fiasco/ and https://imagetragick.com/ come to mind
[cleverdevil] I definitely want "read/unread" for some feeds and definitely not for others.
tantek sounds like /webactions
grant.codes uploaded /File:tweet-deck-channel-filters.png "A screenshot showing how tweetdeck implements filters per "channel" https://indieweb.org/File:tweet-deck-channel-filters.png"
Zegnat Ugh, my token provider tool thing doesn’t work with Quill :( https://github.com/indieweb/indieauth-client-php/issues/16
ancarda Zegnat: If you check Namecheap, they might allow CAA records to be created. You'd need one called "issue letsencrypt.org" (make a new `issue` record per CA you want to whitelist). You can also add "iodef mailto:your@email.address" which will email you when there's an attempted certificate request that was blocked
KartikPrabhu ancarda: it is a wiki. you on't nee to ask permission
markdain.net edited /CAA (+1344) "Added basic information about what CAA is and how to use issue, issuewild, and iodef" (view diff)
petermolnar what is ++ing?
aaronparecki.com edited /Microsub-spec (+3345) "/* Paging */ describe paging mechanism" (view diff)
GWG Like this thing.. http://www.davidghoyle.co.uk/WordPress/?p=1827