#microformats 2022-02-06

2022-02-06 UTC
jacky, Seirdy, edburns[d], Zegnat[d], Christian_Olivie, aspenmayer[d], corenominal[d], hoenir, Murray[d], marksuth[d], MarkJR84[d], shaunix[d], mossymaker[d], hans63us[d], Seb[d], Asaf_Agranat[d], sayanarijit[d], daiyi[d], cygnoir[d], sarahd[d], rattroupe[d], Jeremiah[d], Myst[d], capjamesg[d] and aaronpk[d] joined the channel
#
jacky up to 43 from 34 (from https://chat.indieweb.org/microformats/2022-02-04#t1644017571436900)
#
jacky actually 49!
#
jacky out of 95
#
jacky that's half way :)
ur5us joined the channel
#
jacky so a test failed b/c it's mentioned h-org as a type (the parser does hard failures on unrecognized types)
#
jacky I don't think it's expected to have actually hard failures
#
jacky but I wonder if it's worth noting
#
jacky nvm worked around it
#
jacky gonna pick this up tomorrow; jumped to 75/95 passing (implied stuff is really the bread and butter here)
jacky joined the channel
#
capjamesg[d] Congrats!
#
capjamesg[d] are you building a parser?
[tonz] and jacky joined the channel
#
[KevinMarks] Hm, part of the point of the mf2 changes was not needing to update parsers for new types or fields. Maybe we need to add some tests to reflect that (and also the field and type name filtering rules)
#
sknebel !tell jacky "so a test failed b/c it's mentioned h-org as a type (the parser does hard failures on unrecognized types)" <- there is no concept of "recognized" types in mf2 parsing! don't do that please
#
Loqi Ok, I'll tell them that when I see them next
[jackjamieson] and barnaby joined the channel
#
[tantek] This is why I think h-org is in there, there's no spec for it
dovedozen[d] and P1000[d] joined the channel
#
[KevinMarks] we added the constraints on field names to avoid the utility class false positives, but I don't think we have negative test for those `The "*" for root (and property) class names consists of an optional vendor prefix (series of 1+ number or lowercase a-z characters i.e. [0-9a-z]+, followed by '-'), then one or more '-' separated lowercase a-z words.`
#
[KevinMarks] maybe we need some more abstract parser tests to catch both novel type and property strings, but also have some that should not be parsed (eg my mixed case schema alike ones shouldn't parse now http://www.kevinmarks.com/microformatschema.html )
#
Loqi [Kevin Marks] Microformats 2 and Schema 2015-06-30
#
[KevinMarks] so this used to parse as novel schema like microformats http://www.kevinmarks.com/inourtime.html
#
[KevinMarks] but doesn't now as the types and properties are invalid
#
[KevinMarks] so I'd need to make it no longer camelCase like this to parse http://www.kevinmarks.com/inourtime-compat.html
wackycity[d] and jacky joined the channel
#
Loqi jacky: sknebel left you a message 8 hours, 15 minutes ago: "so a test failed b/c it's mentioned h-org as a type (the parser does hard failures on unrecognized types)" <- there is no concept of "recognized" types in mf2 parsing! don't do that please
#
jacky up to 78 out of 95 tests (the microformat tests are a smaller amount - like 6 less, I have some checks for value class parsing)
[tonz]1 joined the channel
#
jacky ah yeah for parsing, it's more so to support this: https://microformats.org/wiki/microformats2-faq#how_do_you_use_experimental_microformats_and_property_names
#
jacky allowing for a distinction between a vendor extension, experimental and conventional (I guess?)
#
jacky it's mainly decorative for the parser but it helps with serialization from MF2+JSON to the concrete types in Rust (like which ones should have ptd run on it, etc)
ur5us and jamietanna joined the channel
#
jacky lol my parsing of h-recipe keeps thinking that `Fat: 3.4g` is a URL
#
jacky at least when deserializing from json
#
jacky and thus it lowercases the F - making the test break
#
jacky I think I can catch this if I detect a `: ` in the string - a nasty lil' hack tbh
#
[KevinMarks] Did your parser cope with my made up schemas.
#
jacky which one are those?
#
jacky i'm running it against everything in the `microformats_v2` folder
#
jacky https://gitlab.com/maxburon/microformats-parser/-/merge_requests/2 is the PR
KartikPrabhu joined the channel
#
jacky ah did you mean the mixed vendor-prefix ones?
#
jacky is now reading https://github.com/microformats/microformats2-parsing/issues/9
#
jacky lots of good thinking therer
#
jacky *there
#
Loqi [sknebel] #9 "return the normalized absolute URL" for invalid URLs?
#
[KevinMarks] I meant http://www.kevinmarks.com/inourtime.html and http://www.kevinmarks.com/inourtime-compat.html
#
[KevinMarks] From this exercise a while ago http://www.kevinmarks.com/microformatschema.html
#
Loqi [Kevin Marks] Microformats 2 and Schema 2015-06-30
#
[KevinMarks] I was thinking it might be worth making tests to clarify the approved and not versions of type and property syntax
#
jacky right now I'm working on this regex for time strings with merdian suffixing https://regexr.com/6eqmv
#
[KevinMarks] Time is one of those deep complexity problems. Bear's python lib was good at these kind of edge cases
#
[KevinMarks] https://github.com/bear/parsedatetime
#
Loqi [bear] parsedatetime: Parse human-readable date/time strings
#
[KevinMarks] If you don't have a python env to compare the test cases could still be useful
#
jacky I'll check it out (once I get these passing and checked out)
#
jacky this tool has been extremely helpful: https://github.com/mgdm/htmlq
#
jacky for those looking for a cli tool for querying HTML
#
Loqi [mgdm] htmlq: Like jq, but for HTML.
#
jacky agh the need to preserve the time value is understandable but it makes me regret leaning on this library for parsing time values
Osvik[d] joined the channel
#
jacky er, not the time value itself but the original authoring format
#
jacky I guess that's to avoid having the parser do / include time parsing 'magic' that'd come with handling things like timezones
#
jacky agh yeah I'm going to have to chuck all of this datetime wrapping code lol
#
jacky great
#
jacky that actually wasn't so bad - just replacing it for a bunch of regex
#
jacky like four
#
jacky just needing to serialize it correctly since meridan values are a bit of a hint
barnaby joined the channel