#microformats 2015-10-19

2015-10-19 UTC
tantek and fuzzyhorns joined the channel
#
tantek edited /h2vx (+48) "/* issues */ subhead for newest issues" (view diff)
#
tantek edited /h2vx (+212) "/* Fatal error Call-time pass-by-reference */ github issue, noting which line of code is causing the rror" (view diff)
#
tantek edited /h2vx (+11) "/* Fatal error Call-time pass-by-reference */ typo, sign" (view diff)
#
tantek edited /h2vx (+268) "/* issues */ Fatal error Call-time pass-by-reference resolved" (view diff)
fuzzyhorns, MeanderingCode, pfefferle, adactio, behind_you, Erkan_Yilmaz, tantek, joskar, glennjones, Zegnat, eschnou, pfefferl_, TallTed, Soopaman and Hixie joined the channel
#
@Jeremy_Arent RT @a2exandre: Oups... #microformats https://twitter.com/a2exandre/status/648899229373591552/photo/1 (twitter.com/_/status/656152776616103937)
gRegorLove, Soopaman, eschnou and KartikPrabhu joined the channel
#
tantek one big challenge for h-entry is - how do we allow/encourage innovation with / on top of it, without it just becoming a dumping ground for dozens of new properties?
#
aaronpk to avoid needing to add new properties for different kinds of objects, you'd have to have a way for h-entry to have a place to put unknown/new objects
#
aaronpk like how attachments work in email
#
aaronpk for example, I want to publish my bike rides, which have a set of properties like "duration" and "average speed" not to mention the actual route data with all the location points.
#
aaronpk I could define a new object, h-x-route and say that this object has all these properties
#
tantek all microformats allow arbitrary new objects - they just go into the children collection
#
tantek there is no need for new properties for new nested objects
#
aaronpk aha there is children! I kind of forgot about that!
#
tantek children is the trivial modularity extension mechanism built-into all microformats
#
aaronpk maybe that's good enough then
#
tantek thus leaving you to "just" come up with figuring out what your new object needs (if you even need one), and then marking up your content as such
#
aaronpk the post-type-discovery algorithm changes slightly then, right?
#
aaronpk since it would need to look at what is in the children
#
tantek it would when such things become a practice
#
aaronpk consumers of an h-feed that know how to render routes would just need to look for whether there is an h-x-route object as one of the children of the h-entry
#
aaronpk then a reader could show a little map and stuff
#
tantek which then begs the question - does post type discovery itself need to have a core and have modularity for others to extend it?
#
aaronpk hm yeah it would
#
tantek since you're running into a real world itch/example for that, could you add that to https://indiewebcamp.com/post-type-discovery#Issues ?
nitot and eschnou joined the channel
#
aaronpk done
#
tantek thanks!
#
tantek interesting, "feed" was dropped from HTML5's explicit lst of rel values
eschnou joined the channel
#
aaronpk reconsiders his whole internal storage format and thinks about using "children" instead
#
tantek would doing so simplify your storage and your code?
#
aaronpk i think it's about equivalent
#
aaronpk right now i'm storing the route object on a property called "route" internally. when I want to use it, I check "is there a value for the route property" and then use it
#
aaronpk if I switch, the "is there a route" would have to change to iterate over the children and look for an "h-route" object there
#
aaronpk this may be a problem if I want to store more than one object of the same type that mean different things, like a start and end location (contrived example)
#
tantek aaronpk: did you look at / consider any of the previous "directons" research / brainstorming? http://microformats.org/wiki/directions
#
aaronpk not sure that's actually a problem yet
#
aaronpk yeah this isn't directions at all
#
tantek or is "route" different enough from "directions" that you think it deserves different consideration? if so, how are they different?
#
tantek they seem similar
#
aaronpk this is what I mean by a "route" http://aaronparecki.com/metrics/2015/10/19/065451/
#
tantek aaronpk, perhaps start a page http://microformats.org/wiki/route similar to http://microformats.org/wiki/directions, along with stubs for route-examples, route-formats, route-brainstorming?
#
aaronpk also http://aaronparecki.com/metrics/2015/09/01/063757/
#
aaronpk a run with a photo http://aaronparecki.com/metrics/2014/11/06/064017/
#
tantek hmm - the picture looks like directions summary to me
#
tantek as does distance
#
tantek the duration is presumably *actual* instead of *estimated*
#
tantek the method of transport (bike vs walk vs car) is also typically published in directions (e.g. google maps)
#
aaronpk it's not really directions summary since it's actual location traveled rather than planned, and i would also never describe the route in words used to describe directions
#
tantek seems like the text equivalent for the image could describe the route in words similar to those used in directions
#
tantek and would be useful for accessibility, search indexing etc.
#
tantek tries to avoid "would be nice" :)
#
aaronpk also the data I actually have is a bunch of lat/lng, not words
#
tantek "head east south east from ???,??? lat long"
#
tantek "turn left at ???,??? lat long"
#
tantek etc. :)
#
aaronpk but that's not the data i have
#
tantek sure looks like it from the map
#
aaronpk no it's literally a bunch of lat/lng locations
#
tantek with a start and stop
#
aaronpk and timestamps for each location
#
tantek that seems like enough to render such text as a textual summary
#
aaronpk it's not even "turn left at..." it would be "turned left at ..." since it happened in the past and is not an instruction
#
tantek yes!
#
aaronpk turning that into text is harder than you'd think
#
aaronpk also would likely be completely unusuable in text form
#
tantek I'll take your word for the difficulty.
#
aaronpk am i going to have to actually do this to prove a point?
#
tantek no, I realize that geocoding from lat long to streetnames is a non-trivial problem that people have to call pay-for APIs to do :/
#
tantek was just hoping you'd already figured that out somehow
#
aaronpk ever since simplegeo shut down i haven't found a good lat/lng to street intersection api
#
tantek given being at ESRI
#
tantek dang
#
aaronpk geonames is close, (us only) but isn't as good as simplegeo
#
tantek what about open street maps? tommorris ?
#
aaronpk ironically simplegeo built that API because I asked them to
#
aaronpk geonames uses OSM data
nitot joined the channel
#
tantek ok, for now can you at least start http://microformats.org/wiki/route-examples and cite your published examples that you linked above?
#
aaronpk even if there were a good API for that, there's still the problem of describing the route as I run through the park where there are no street names and the path wanders around
#
tommorris tantek: OSM has Nominatim for reverse geocoding and there are open source routing engines on top of OSM data (have seen them support cars, cycles, walking, even horse riding)
#
aaronpk is "trip" a better name for it?
#
tantek or is a route an optional object of a trip?
#
aaronpk i've never been quite satisfied with the term "route"
eschnou joined the channel
#
tantek thinking of the dopplr use of "trip"
#
aaronpk i do have trips without routes (many plane flights)
#
tantek route makes sense in the way you're using it with all the coordinates
#
tommorris As for a route through a park: you are welcome to add footpaths to OSM :) and just take a GPX and turn it into geoJSON and render that
#
tantek right!
#
tommorris If the GPX is too large, run a simplification algorithm over it
#
aaronpk i just think a route doesn't make sense to have timing information necessarily, nor does it convey that it is a thing that happened at a single point in the past
#
tommorris Class object distinction between the route and the individual run/walk/cycle through it?
#
tantek you can brainstorm the name / bikeshed later too - ok to have a page like route-examples and then come up with h-somethingelse later
#
tantek tommorris: hence why I thought of directions
#
tantek since directions clearly have a route
#
tommorris So directions would be defined by the fact it is ordered?
#
tommorris An ordered list almost... <ol> :)
#
tommorris <ol>é! <ol>é!
#
tantek tommorris: see and feel free to expand upon http://microformats.org/wiki/directions
#
tommorris I kinda have to go and work dinner off by engaging in directed travel along a route...
rhiaro, nitot, fuzzyhorns, KartikPrabhu and andicascadesf joined the channel