#ZegnatThe second part will also help you set-up vhost, which means you can have development domains (e.g. jeremycherfas.dev) instead of subfolders in localhost.
#snarfedi still don't fully grok the non-feed-reader use cases and context...but would this fit that too? ie only render u-photo in atom content if reader=true?
#snarfedguess i'm struggling with how to map the obvious human judgments to an algorithm :P
#aaronpkyou can drop the location property and the post still means more or less the same thing to poeple reading it. but if you drop the photo, the post loses most of its context
#snarfedi definitely get that human judgment. i just don't yet see a corollary for it in any mf2 property, parsing rule, or other algorithm that we currently have
#aaronpki think there is probably a short list of mf2 properties that are considered content of a post
#Loqi[snarfed] narrowed down the thinking here a bit [in an IRC discussion w/@aaronpk today](https://chat.indieweb.org/dev/2017-08-11#t1502474343806000).
some mf2 properties are primarily _data_. `content` and `name` obviously, but also `photo`, as discussed her...
#ZegnatNot long ago sknebel and I were looking at mf photo property and post type discovery, wether it could make mistakes or not, with photo being an implied property. Though I don’t think that comes in here either?
#ZegnatAh, yeah, I think that was it. If photo was implied it could have consequences (adding u-photo to u-feature etc) but in reality the implied rules are never applied.
#sknebelstill leaves the question if they are useful then, and if everyone uses complicated enough layout. really need a serverfarm to run random experiments over indiemap data ;)
#sknebel(been trying today to download all Let's encrypt certificates ever. need to free disk space for that as well...)
#sknebel(correction: not all ever, all in CT logs)
#snarfedsknebel: try indiemap? bigquery + regexps and jsonpath let you do some pretty sophisticated queries over the parsed mf2 and raw json, respectively
#snarfedfree up to 1T queried, then cheap after that
#sknebelbeen playing with it, haven't quite grokked how to use bigquery effectively though
#snarfedok. i don't know much jsonpath myself, but otherwise i'm happy to help
#Zegnatsnarfed but access to the original HTML? That’s the only way to see if implied u-photo rules were applied (test: is there photo on parsed mf2 but missing as class in html?)
#sknebelok, processing html is $1 per query unless I get it right in 4 tries ;)
#snarfedfortunately it doesn't charge you for queries that fail (for any reason)
#snarfedalso, generate a small view of the pages table, e.g. limit 10000, and save to a new table test on that, will be way less data queried. then only run on the full table once it's working.