#dev 2018-11-18

2018-11-18 UTC
#
[tantek]
GWG are you using / following anything proposed here? https://indieweb.org/h-feed#feed_autodiscovery
#
GWG
rel-feed
#
GWG
I did the work before some of that was added.
#
[tantek]
!tell Zegnat stub section for you 🙂 https://indieweb.org/h-feed#Zegnat_proposal_2018
#
Loqi
Ok, I'll tell them that when I see them next
#
GWG
I am looking at the fact that a feed that is not the URL of the current page is treated as a full feed.
#
GWG
That confuses me
#
GWG
In your tantek example, I actually did what you suggest...treat the first feed as the primary feed.
#
[tantek]
What do you mean "full feed"?
#
GWG
"Alternatively, if an h-feed has a u-url u-uid property that is not the URL of the current page itself, then that u-url u-uid URL can be treated as the canonical full feed. "
#
GWG
It says it there.
#
[tantek]
I'm not sure the term "full feed" is useful
#
[tantek]
GWG, are you needing an explicit url/uid for the specific feed? Because you don't have access to the HTML ID of the h-feed element while you're doing your feed discovery?
#
jacky
b qen
#
GWG
[tantek]: PHP-MF2 doesn't show it, which is what I use to parse microformats
[grantcodes] joined the channel
#
[grantcodes]
Doing some together updates and it's now got dark mode maps. Looks so good 🌚
#
GWG
[grantcodes]: It what?
#
GWG
[tantek]: So, maybe it is a issue to file on the parser.
#
sknebel
will write a proposal tomorrow
#
GWG
I checked. None of the parsers set a uid or url based on that.
#
GWG
It would certainly be helpful.
#
GWG
sknebel: Where do we keep those proposals?
#
jgmac1106
okay added the author h-card but threw it and publication date info in a no display on homesite but not on article page
#
GWG
[grantcodes]: I meant for what software
#
[tantek]
Interesting. ID attribute could be useful to capture on h-* objects
#
GWG
[tantek]: I think so, when there are multiple in a page.
#
[grantcodes]
Wha? It's in together. There's been a dark mode toggle for ages
#
[tantek]
and/or perhaps the full URL with the ID, just as we compute absolute URLs for relative URLs in properties
#
GWG
[grantcodes]: I thought so.
#
GWG
[tantek]: I would suggest the full URL with the ID as the uid.
#
[grantcodes]
It has also supported system settings based dark mode in edge, way before the current osx dark mode excitement 😛 I guess I should implement that as well though
#
[tantek]
GWG, that may be a way to imply a u-uid in the absence of any explicit u-uid
#
GWG
[tantek]: Promising .
#
[tantek]
On one hand I'm not sure if that's something to rely on. On the other hand "ID" is very well defined in HTML to supposed to be unique for the page which maps well to a uid based on the URL, and may be "good enough" to solve these use-cases.
#
GWG
But using it as a fragment identifier is established
#
[tantek]
!tell sknebel perhaps capture some of these ideas (capture ID on h-* element explicitly on the item, perhaps *also* use it to generate an implied u-uid with the URL of the page in the absence of any explicit u-uid on the h-*) in the issue you are filing / proposal for microformats2-parsing/issues
#
Loqi
Ok, I'll tell them that when I see them next
#
GWG
So, if there isn't a u-uid on a h-* element, the URL of the page, plus the id as a fragment...
#
[tantek]
GWG, yes, that's a good fallback, and perhaps you can try to use that in your code to see if it works for you
#
[tantek]
confirming whether or not an algorithm works for the intended use-case is very useful
#
[tantek]
GWG, are you able to get the ID attribute from the HTML?
#
GWG
[tantek]: I'd have to go through and search for it. The problem I have to think about is this. How do I match the return from php-mf2 with a second run?
#
[tantek]
yes that seems tricky / fragile
#
[tantek]
I suppose I could add explicit rel=feed in the head that would point to each h-feed?
#
GWG
I'd rather see if we could get it behind a flag inside php-mf2 that I could experiment with rather than building a php-mf2 substitute.
#
[tantek]
would that help?
#
[tantek]
makes sense
#
GWG
[tantek]: I may write the php-mf2 issue now.
#
Loqi
[dshanske] #206 Add experimental implied UID for h-* elements
[jgmac1106] joined the channel
#
[jgmac1106]
Yeah! It works. Though now I see why I need to add a photo to the h-card of each h-entry
#
[tantek]
GWG++ thanks!
#
Loqi
GWG has 42 karma in this channel over the last year (169 in all channels)
#
[tantek]
!tell sknebel when you're writing up that mf2 parsing proposal see also related https://github.com/microformats/php-mf2/issues/206
#
Loqi
Ok, I'll tell them that when I see them next
#
Loqi
[dshanske] #206 Add experimental implied UID for h-* elements
#
[jgmac1106]
Link is missing, no where to click. I need to fix that
#
jgmac1106
forgot u-url
eli_oat, [jgmac1106] and [eddie] joined the channel
#
[eddie]
tantek regarding the conversations around people centric focus over feeds, how do we address the issue of people having various feeds? For example, I don’t want to list all of my feeds on my homepage, but if we want someone to focus on following “me” I should be able to “advertise” my available feeds from my h-card
#
[eddie]
similarly Zegnat’s h-card has NO feeds on the same domain. So how would I follow Zegnat in this line of thinking? I definitely think that’s a direction to head to, but there is a missing technical barrier between me entering Zegnat’s url into a “follow” box and me being able to know what feeds Zegnat might make available
#
gRegorLove
Interesting implied uid idea. I like it.
[tantek] joined the channel
#
[tantek]
Eddie not sure I understand if you mean one use case or several?
#
[tantek]
“People having various feeds” is exactly what a person-centric approach here solves and makes easier to maintain over time
#
[tantek]
If someone gives you a url without any feeds then they are indicating they don’t have anything for you to follow
#
[tantek]
Anything else like multiple domains is just like someone having multiple accounts on a silo - that’s a social issue, not s technical issue
tantek, AngeloGladding, globbot, iasai and [eddie] joined the channel
#
[eddie]
hmm I think you misunderstand. So my identity is eddiehinkle.com but I have different feeds: a timeline feed (eddiehinkle.com/timeline), articles feed (/articles), work feed (/work). How would someone given my Url be able to be given those options: “Timeline, Articles, Work,etc”
#
[eddie]
I shouldn’t have to display the content of the feeds themselves on my homepage in order for those to be discoverable if someone is attempting to “follow me” rather than “follow a feed”
#
[eddie]
However in the current model if you want to follow my articles, you have to go to eddiehinkle.com/articles/ and follow that url
gRegorLove and [tantek] joined the channel
#
[tantek]
Pretty sure /h-feed page and a bunch of the proposals already cover that with rel=feed
#
[tantek]
link tags in the head to feeds on other pages
#
gRegorLove
Ran into indieweify.me failing a rel-me check because one page is XML: https://github.com/indieweb/indiewebify-me/issues/78
#
Loqi
[mro] #78 false negative testing https://try.gogs.io/issue-5008
#
gRegorLove
I commented that php-mf2 doesn't parse XML documents currently, but then checked and saw Python and Go parsers do
barpthewire joined the channel
#
[eddie]
It’s just an interesting change of posture because I feel like linking to various feeds from a single page has been discouraged in the past
#
[eddie]
but I do think it’s useful in this case of focusing on people.
#
[eddie]
so you would just link to each feed you want to publicize as rel=feed then the user could select which “topics” they are interested in
#
[eddie]
Hmm nope, sure enough that almost exact use case is detailed on /h-feed. I wonder why I ended up with that misconception that it was frowned upon
#
[eddie]
That’s definitely a solution, rel=feed all the main feed options for users from the homepage with the title providing user friendly explanations of feed content
#
petermolnar
!tell aaronpk is there any api or import way to populate aperture from an extermal source with subscriptions?
#
Loqi
Ok, I'll tell them that when I see them next
iasai joined the channel
#
sknebel
petermolnar: https://github.com/pstuifzand/ekster can do that as a client
#
Loqi
sknebel: [tantek] left you a message 10 hours ago: perhaps capture some of these ideas (capture ID on h-* element explicitly on the item, perhaps *also* use it to generate an implied u-uid with the URL of the page in the absence of any explicit u-uid on the h-*) in the issue you are filing / proposal for microformats2-parsing/issues
#
Loqi
sknebel: [tantek] left you a message 9 hours, 39 minutes ago: when you're writing up that mf2 parsing proposal see also related https://github.com/microformats/php-mf2/issues/206
#
Loqi
[pstuifzand] ekster: microsub server
[jgmac1106] joined the channel
#
[jgmac1106]
!tell Eddie I am experimenting with putting feeds in my h-card
#
Loqi
Ok, I'll tell them that when I see them next
jgmac1106 and jeremych_ joined the channel
#
petermolnar
sknebel: thank you very much for the tip, so far, it doesn't work: 2018/11/18 12:02:00 main.go:204: invalid character 'm' looking for beginning of value
#
petermolnar
when I try to auth `ek`
[jgmac1106] joined the channel
[kevinmarks] joined the channel
#
[kevinmarks]
Implied uid on h-entry with page url is going to be confusing
#
jgmac1106
I think it wasn’t in realtion to h* but to h-feed in general but I could be wrong
#
GWG
kevinmarks, page url and html id tag... but how so?
#
jgmac1106
@gwg was having trouble parsing the three different feeds on tantek.com and folks were brainstorming ways uid could make each recognizable
#
jgmac1106
speak of the devil
#
GWG
Yes, but it could apply to any page with multiple properties
#
GWG
How do I refer to just one of them?
#
jgmac1106
why can’t you use the p-name of the h-entry to tell them apart/
#
jgmac1106
I mean h-feed
#
GWG
jgmac1106, I want to pass a single thing
#
GWG
So, if I could pass http://tantek.com#recentarticles instead of http://tantek.com and "Recent Articles" it is easier for most people. You can have a link to the feed that way
#
Loqi
Tantek Çelik
#
jgmac1106
as post types become more popular maybe h-feeds need something like p-type=“articles”
#
jgmac1106
I am redoing my following page so everyone is an h-card: https://jgmac1106homepage.glitch.me/following.html [gwg] [kevinmarks] would you make each channel it’s own h-feed or would it be better to group in /collections or simply nothing at all?
#
GWG
jgmac1106: I don't think that is necessary.
#
GWG
The p-type thing
#
Loqi
[sknebel] #44 parse HTML id= property
#
jgmac1106
just strategizing, seems to me if there would be way to p-name of h-entry automatically into tantek.com/#RecentArticles that might work
#
sknebel
I agree with kevinmarks that implying uid is not a good idea
#
sknebel
reasons in the proposal
#
sknebel
you can still use url with fragment to store the "address" of the feed
#
jgmac1106
[gwg] part of the reason I am experimenting with putting feeds in my h-card and then only having one h-feed of recent articles on my homepage
#
GWG
jgmac1106: Links to feeds, I'm assuming, not the actual feed
#
jgmac1106
People need cues on feeds to follow. I wouldn’t know to follow eddiehinkle.com/timeline if he didn’t tell me…yes I put the actual feeds and people told me to just use a regular link with rel=“feed”
#
jgmac1106
muat say it felt kinda of cool to show up in aperture after adding juts a few properties
#
[kevinmarks]
A lot of blogging systems put id on the posts - historically that was what blogs looked like - daily/weekly pages with posts on, permalinks with id (or name if you go back far enough)
#
[kevinmarks]
So if you put h-entry on the post container with the id, you will need to be careful
#
GWG
This is why I wanted to discuss this
#
sknebel
does my proposal make sense?
iasai joined the channel
#
jgmac1106
sven sorry for noob question but would it be like <a rel=“feed” id=“RecentArtcles”>
#
sknebel
no, this got nothing to do with linking to external feeds
[jgmac1106] joined the channel
#
jgmac1106
glad I asked…oh yeah wouldn’t need that for h-feed…it is the feed
#
sknebel
if you got multiple h-feeds in the page, one of them could e.g. be <div class="h-feed" id="recentArticles">..., and e.g. a reader could then remember that the user subscribed specifically to the feed with the id "recentArticles"
#
sknebel
if the feeds are external and have individual urls, you don't need that
#
jgmac1106
i think that makes sense
#
jgmac1106
[sknebel] why does <div class="h-feed" id="recentArticles"> give you more info than <div class="h-feed" p-name=“recentArticles">
#
jgmac1106
couldn’t parser logic be if h-entry has p-name “remember that the user subscribed specifically to the feed"
[tantek] joined the channel
#
[tantek]
tries to disentangle scrolback. Sometimes chat does make threading convos difficult, even on one “topic” with branching subtopics
#
jgmac1106
or is it because id makes #recetnArticles on a url?
#
Loqi
[sknebel] #44 parse HTML id= property
#
[tantek]
sknebel++ thanks for writing this up!
#
Loqi
sknebel has 30 karma in this channel over the last year (84 in all channels)
#
[tantek]
Excellent analysis and makes sense to me, both on adding “id”, and (for now) not implying u-uid
#
[tantek]
What that does is leave it up to the consuming code to use (or not) the “is” property only as necessary.
#
[tantek]
E.g. for the page multifeed case, the consuming code can combine the canonical URL (per HTTP redirect semantics as noted in the issue) and construct a fragment reference only if needed to one h-feed when there are more than one in the page
#
sknebel
exactly. it can still work with fragment urls, but has more knowledge on how to build the right one
#
sknebel
also doesn't break if the author later adds a u-uid because they added a permalink for the feed
#
[tantek]
Right, there’s always a challenge in finding the right balance of where to put the “smarts”, in the parser for all microformats, or in the code for specific consuming use-cases
#
jgmac1106
what is a fragment url?
#
Loqi
It looks like we don't have a page for "fragment url" yet. Would you like to create it? (Or just say "fragment url is ____", a sentence describing the term)
#
[tantek]
What is a fragment
#
Loqi
A fragment is part of a URL, and typically refers to a fragment identifier at the end of a URL preceded by a # but may also refer to a media fragment which may use a fragment identifier or a query string https://indieweb.org/fragment
#
[tantek]
fragment URL is a URL with a [[fragment]] identifier.
#
sknebel
jgmac1106: regarding benefit over using p-name, the id is by design considered unique on the page, and one might consider the p-name to be less stable(?). for only the feed scenario the p-name thing might work (although see GWGs argument for making shareable URLs), but you'd loose the other use case
#
[tantek]
GWG, do you think your code would work with just the “id” property on a microformats object? (I.e. no implied u-uid)
#
[tantek]
If so, go ahead and update the phpmf2 issue you created to only ask for that per the mf2 parsing issue 44 linked above
#
[tantek]
I think the guidance there from a feed discovery algorithm perspective is to use that “id” if present is to have your code synthesize a “u-url” for the h-feed with that id as fragment
[kevinmarks] joined the channel
#
[kevinmarks]
You could explicitly add a u-uid to yours, Tantek
#
sknebel
added a proposal for the specific spec change language too
#
GWG
It probably would
[cleverdevil] and [tantek] joined the channel
#
[tantek]
Kevinmarks the point is to not have to add more burden to publishers
#
[tantek]
Adding an ID attribute should be sufficient
jgmac1106 and [kevinmarks] joined the channel
#
[kevinmarks]
Right, but each h-feed having a u-url with a fragment to the id makes sense anyway.
[jgmac1106] and [eddie] joined the channel
#
Loqi
[eddie]: [jgmac1106] left you a message 6 hours, 26 minutes ago: I am experimenting with putting feeds in my h-card
#
[eddie]
What’s that, you say? h-cards being returned from a Micropub endpoint query?
#
[eddie]
🙌 🙌
#
[jgmac1106]
eddie trying to think the best way to display a ton of h-cards grouped into channels. What is your thinking? Currently just using ul and an h3 for the group name
[Csongor] joined the channel
#
[eddie]
I personally wouldn’t bother displaying channels publically
#
[eddie]
For me that’s an information organization tactic as opposed to a public relationship to display
#
[eddie]
but if your channels are more about groups of people, relationships and organizations
#
[eddie]
then it could be h-cards with children h-cards
#
[eddie]
but if it’s just information management, yeah a list is probably the cleanest
#
[eddie]
If you feel like you need to organize them because there are so many I might lean more towards exposing your relationship to these people through groups rather than by showing your channels
#
[eddie]
These h-cards I’m currently doing are for an auto-complete UI
#
[eddie]
so there’s no grouping, just an ordered list (by nickname)
#
[eddie]
I’ll also probably try to avoid grouping as long as possible, but if it becomes not possible likely group by life relation
#
Zegnat
Loqi, any messages for me?
#
Loqi
Zegnat: petermolnar left you a message 20 hours, 7 minutes ago: reading up on webkit vs search box vs CSS, who on earth came up with this idea?!
#
Loqi
Zegnat: [tantek] left you a message 17 hours, 12 minutes ago: stub section for you 🙂 https://indieweb.org/h-feed#Zegnat_proposal_2018
#
Zegnat
That’s what I thought, hehe :D Caught up on this channel. Interesting discussions on feeds and nickname caches!
#
GWG
Your thoughts?
#
Zegnat
My thoughts on the feeds thing: we still need to try and document a discovery flow so implementations do not diverge too much. I have no thoughts on nickname caches atm, as I do not have one of my own in the pipeline.
#
Zegnat
I will point out that my “proposal” was purely the flow I saw for tantek.com; so e.g. comments made about whether nesting inside h-card was important weren’t something I was trying to tackle at all
#
Zegnat
tantek++ for documenting my chat spew on the wiki
#
Loqi
tantek has 27 karma in this channel over the last year (82 in all channels)
#
GWG
I don't either, only in my imagination
#
Loqi
[sknebel] #207 Implement HTML id parsing proposal
[kevinmarks], j12t_ and j12t joined the channel
#
jacky
yo so https://indieweb.org/Micropub-extensions#Query_for_Post_List i've added support for "limit" and "type" in my implementation (Koype)
#
jacky
I _think_ that's about it but maybe some people considered having a "search"/"filter" kind of thing?
#
jacky
might be overloading that particular value but eh
[tantek] and wagle joined the channel
#
jgmac1106
Could I do this if I have mutliple ratings in one h-review <p class=“p-category>Domain 1<span class “p-rating” value=“4”> Exemplary</span></p>
#
jgmac1106
or would each scale be considered a p-item..going to play a bit with this
#
petermolnar
what is search?
#
Loqi
search in the context of the IndieWeb refers to being able to search your personal site for your own content https://indieweb.org/search
#
Zegnat
jgmac1106, I am not sure what you mean. But if you have multiple p-rating on the same h-review there is probably no good way to separate them in the mf parser output
#
jgmac1106
Trying to show a rubric that has an overall rating, then domain ratings, and possibly even a third layer. speaking psychometirically each scale in a rubric is an item so p-item makes sense in at that level, going a third level down didn’t know if p-note or p-category would be better
#
jgmac1106
basically nested ratings somehow
#
[kevinmarks]
hm. hReview (old version) had ratings for tags. h-review doesn't specify these. - [tantek], was that deliberate?
#
Zegnat
“wouldn’t work” ... you tell me jgmac1106, is this an expected or useful output? http://php.microformats.io/?id=20181118194148301
#
jgmac1106
yeah trying to think about how we could set up a publish from your own site journal, need a system for syndicated peer review
#
Zegnat
“a system for syndicated peer review” - isn’t that just people sending you a comment?
#
jgmac1106
yes it would be, but need to put fancy numbers behind the comments to justify “rigor” of review
#
Loqi
[Greg McVerry] workflow I was thinking Journal website puts out call you write draft, reply to call open peer editors use journal or own website they each write h-review as reply to draft edits and revisions back and forth as replies final draft syndicated to...
wagle joined the channel
#
jgmac1106
is kind of what I am thking.
#
jgmac1106
but no that is useless parsing
#
[kevinmarks]
http://microformats.org/wiki/hReview#Multidimensional_Restaurant_Review is an example of multiple ratings, but not sure if any current parsers support that.
#
Loqi
hReview 0.4 (in progress)
KartikPrabhu and [jgmac1106] joined the channel
#
[jgmac1106]
haven't seen rel="tag" is sometime, but yeah something like that
#
[kevinmarks]
for mf2 to parse it you'd need to make rating an h-rating
#
jgmac1106
I already get the data out in a way that lets me consume it in a way I needed, just wondered if I could add microformats to a journal review and then possibly into ReVIEW itself https://quickthoughts.jgregorymcverry.com/2018/11/18/indieweb-question-on-h-review
snarfed joined the channel
#
jgmac1106
but for proof on concept I will build a single holistic rubric on a five point scal and use the h-review in Known
wagle joined the channel
#
[jgmac1106]
yes, now were looking like the json files I currently export....wondering if a json to mf2 would be easier than redoing everyone of my views and templates
wagle and [tantek] joined the channel
#
[tantek]
Kevinmarks in the conversion to mf2 we flattened and dropped lots of things
tantek, eli_oat and [Vincent] joined the channel
#
[kevinmarks]
sounds like JG has a use case
tantek_ and snarfed joined the channel
#
[jgmac1106]
I Can make a single holistic scale rubric for Academic reviews for now, I do not have a parser in ReVIEW so even I would not be consuming them yet.
#
[jgmac1106]
The way the old hreview parsed would be what I need if I did ever start to really use mf2 with portfolio review
iasai joined the channel
#
[kevinmarks]
nesting is explicit rather than implict in mf2, so you would need to do h-rating like I did in that example.
#
[jgmac1106]
Yeah I will try and mock up a template. What you did would work
eli_oat1, mblaney and [tantek] joined the channel