#microformats 2020-05-11

2020-05-11 UTC
fkrt2 joined the channel
#
aaronpk
it annoys me every time that the h-card property name is country-name instead of just country like the rest of them
KartikPrabhu joined the channel
#
[tantek]
we took it from vCard which I believe made that distinction because normally a "country" field in various formats is expected to be a country short code
#
aaronpk
ah interesting
#
fkrt2
Hello. Why http://microformats.org/ does not have a SSL/TSL certificate?
[chrisaldrich], beko, gRegorLove__ and cjw6k_ joined the channel
#
[tantek]
It's been an issue with the hosting provider, and we literally have not had the volunteer time/expertise to transfer to a new hosting provider
Aishawhite, KartikPrabhu, [fluffy], jamietanna, reidab_, aaronpk, bear, dckc, [Rose], [LewisCowles], [Sadik_Shahadu], [jgmac1106] and indy joined the channel
#
sknebel
!tell tantek: If it would help you with the rel-registry move to convert the table to GH markdown or something else I could do in one concentrated sitting I could do that
#
Loqi
Ok, I'll tell them that when I see them next
#
[jgmac1106]
gwg tantek I could see haudio for my audiopodcasts and poems as well, I haven't given each show a unique url (well a link to a text of a poem) but I could give them an id.
#
[jgmac1106]
I want to be able to send someone specifically a link to a poem or podcast i just them the link to the audio file and not the h-entry
[KevinMarks], [prtksxna], fkrt2, [grantcodes] and [aimee] joined the channel
#
[aimee]
Hey, just thought I would introduce myself 👋 - and I’ve just opened a PR against the test suite to update the references to NPM and dependencies: https://github.com/microformats/tests/pull/118
#
Loqi
[aimee-gm] #118 Correct dependencies in package.json and remove NPM reference
#
[KevinMarks]
would it make sense to republish on npm?
Sajesajama and fkrt2 joined the channel
#
sknebel
not important IMHO
#
[aimee]
I think it only makes sense to republish on npm if we intend to keep it up to date
fkrt2 and [jgarber] joined the channel
#
[jgarber]
[aimee] Welcome! I was just looking at your PR.
#
[jgarber]
I’ve been thinking about the npm package aspect of microformats/tests, so thank you for re-sparking that.
#
[jgarber]
It’s unclear who, if anyone, is using microformats/tests as an npm package. Though, it looks like [aimee] pulled it into a project:
#
[jgarber]
[jansauer]’s project (https://github.com/jansauer/htmlparser2-microformat) doesn’t appear to be using the npm package but instead uses the repository as a Git submodule.
#
[jgarber]
…and [Glenn_Jones]’s project (https://github.com/glennjones/microformats-testrunner) pulls the package from a tarball at a GitHub URL.
#
Loqi
[glennjones] microformats-testrunner: A microformat 2 test runner
[jansauer] joined the channel
#
[jansauer]
Yes. taking the test straight from the source 😄
#
[jgarber]
[jansauer] I’m doing the same in a Ruby library.
#
[jansauer]
[jgarber] nice, always wanted to add this to the php lib
#
[jgarber]
So, building on [aimee]’s point… we could:
#
[jgarber]
1. remove the npm code from microformats/test (since it appears no one is using the project that way), or
#
[jgarber]
2. go all-in and provide installable packages for as many languages as possible.
#
[jgarber]
If we go route number one, we should probably figure out how to publish a new version to npm with a deprecation notice.
#
[jgarber]
(that the package is deprecated, that is, not that the test suite is deprecated)
#
[aimee]
I’ve listed my dependency as a github dependency - as it’s a dev dependency, this doesn’t bother me
#
[aimee]
(And it’s nice to always have the latest tests from the library)
#
[jgarber]
…and if we go with number one, that _does not_ preclude us from building projects that package the test suite for specific languages.
#
[jgarber]
e.g. `microformats-test-suite-ruby`, `microformats-test-suite-node` , etc.
#
Zegnat
PHP is also taking the tests straight from the source, if that counts for anything. The package as listed on packagist is out of date
#
[jgarber]
[aimee] Agreed! Pinning to a commit SHA direct from the repo is what I’ve been doing, too.
#
[jgarber]
…since we aren’t versioning the test suite with tagged releases.
#
[jgarber]
☝ Which is _another_ thought I’d been noodling on.
#
[aimee]
I would only suggest going the route of maintaining packages in the various package managers if they were automatically published in ci
#
[jgarber]
_great_ point
#
Zegnat
According to packagist, it is automatically being updated, but clearly it is not there :( So I also had to pin a commit: https://github.com/microformats/php-mf2/pull/163/files#diff-b5d0ee8c97c7abd7e3fa29b9a27d1780R19
#
[jgarber]
Path of least resistance then is… remove the npm-related code and documentation and encourage reliant parties to pull the code into their project as they see fit (e.g. submodule, copy/pasta)
#
[jgarber]
☝ That’s my proposal
#
[aimee]
And I’d like to switch the dependencies to devDependencies, as otherwise npm pulls them in on an install from github, I’ve had to fork to stop this
#
[jgarber]
If we removed `package.json` , etc. from microformats/tests, that wouldn’t be an issue.
#
sknebel
yeah. don't think the repo should have package stuff in it
#
[jgarber]
You’d add microformats/tests to your project as a Git submodule: https://git-scm.com/book/en/v2/Git-Tools-Submodules
#
[aimee]
To remove package.json you would also have to remove the local server
#
[aimee]
Not sure if anyone uses that? Or even if it works?
#
[jgarber]
Oh, that’d be great, too. Okay, this is growing beyond the scope of your PR. 😄
#
[jgarber]
> Not sure if anyone uses that? Or even if it works?
#
[jgarber]
Great questions. 😄
#
[aimee]
> Oh, that’d be great, too. Okay, this is growing beyond the scope of your PR. 😄
#
[aimee]
It is a little 😉
[tantek] joined the channel
#
[jgarber]
[aimee] I have one piece of feedback I’ll add to your PR. We can review that and then I’ll spin up an Issue to discuss the larger issue of what we do with the package-related code for the test suite.
#
[aimee]
Sure
[Sadik_Shahadu] and fkrt2 joined the channel
#
[jgarber]
[aimee] Review submitted!
#
[jgarber]
Aimee++ for kicking off the conversation about the test suite 🎉
#
Loqi
Aimee has 1 karma over the last year
[LewisCowles], [tw2113], [jgmac1106] and rodolfojcj joined the channel
#
[KevinMarks]
I think a lot of this is left over from Glenn Jones's original multi-implementation tester.
#
[jgarber]
Oh, absolutely.
#
[jgarber]
I’m working on an Issue on GitHub right now to work through some ideas.
#
[KevinMarks]
iirc part of the point of a local server was to allow testing of relative urls
#
[KevinMarks]
ie without a HEAD in the test
#
[KevinMarks]
but serving them from a known domain would do that too
#
[jgarber]
Okay, here we go. microformats/tests idea tracking issue: https://github.com/microformats/tests/issues/119
#
Loqi
[jgarber623] #119 Discussion: Project/Repository Ideas
#
[jgarber]
I’m v. interested in getting everyone’s thoughts there.
#
[tantek]
sknebel, that would help a lot yes.
#
Loqi
[tantek]: sknebel left you a message 4 hours, 30 minutes ago: If it would help you with the rel-registry move to convert the table to GH markdown or something else I could do in one concentrated sitting I could do that
#
[tantek]
Beforehand though I think we need to figure out how to represent the multiple tables in git. Separate files? Or multiple tables in one file? Etc. suggestions welcome!
#
sknebel
[tantek]: I think just transforming the wiki page close to as is into a single file in a repo is fine as a first step. If that doesn't work out, splitting it up can still be done, but I don't see an obvious reason why its not an option and some progress is better than none
#
[KevinMarks]
is the markdown table syntax clear enough?
#
Loqi
[@simonw] TIL you can copy and paste any HTML table into a GitHub issue comment and GitHub will automatically convert the HTML into Markdown table syntax for you! https://pbs.twimg.com/tweet_video_thumb/EXlt2BqUMAEH4jR.jpg
#
[jgarber]
[aimee] I’ll rebase and merge that microformats/tests PR of yours. Sound good?
#
[aimee]
Yep!
#
[jgarber]
…and done!
fkrt2 joined the channel
#
[KevinMarks]
Datasette is neat.
#
[tantek]
[aimee]++ thanks for the PR!
#
Loqi
[aimee] has 2 karma over the last year
dougbeal|mb1, [chrisaldrich], [jgmac1106], [jeremycherfas], SpencerDub, gRegorLove, [fluffy], fkrt2, [snarfed], KartikPrabhu, rodolfojcj and [jgarber] joined the channel
#
[jgarber]
Does anyone have any historical context for the `rels` and `rel-urls` structures in the microformats2 parsing spec?
#
[jgarber]
Curious as to how they took form, how parsers/consumers are using the results, etc.
#
[jgarber]
I’ve got a microformats2-parsing issue related to `rels` / `rel-urls` and was hoping to get a bit of background before posting.
[KevinMarks] joined the channel
#
[KevinMarks]
I think the discussion is on the wiki. The point of rels was to collate the links by type, so you could find all the rel=me or whatever. The point of rel-urls was to find all the rels that apply to a single link without having to walk all the rels. As the parser is walking all of the links with rels anyway, it can build these and save the consumer time and effort.
#
[KevinMarks]
This makes sense for xfn where you potentially have multiple relationships
#
[jgarber]
Copy that. I assumed something in that vicinity was the background.
#
[KevinMarks]
Can also be useful for mf1 fallback where rels are used more
#
[KevinMarks]
Though that often has containment rules
#
[jgarber]
The value of the `rels` hash is super obvious. `rel-urls` less so with the current implementation. This is where I hit an inconsistency btw. implementations that I think is owing to some confusion in the spec.
#
[jgarber]
`rel-urls` are keyed by unique URL, no problem. The `rels` key within that is an array that’s concatenated with new values as their found, no problem.
#
[jgarber]
The remaining keys in each `rel-urls` URL’s value are… strings.
#
[jgarber]
`media`, `hreflang`, `text`, etc.
#
[jgarber]
…which I can’t account for in the spec.
#
[jgarber]
Parser results vary pretty wildly when given some HTML like:
#
[jgarber]
<a rel="me" href="https://sixtwothree.org">Jason Garber</a>
#
[jgarber]
```<link rel="me" href="https://sixtwothree.org">
#
[jgarber]
<a rel="home" href="https://sixtwothree.org">Go back home</a>```
#
Loqi
Jason Garber
#
Loqi
Jason Garber
#
Loqi
Jason Garber
#
[KevinMarks]
Which are mostly for rel=alternate
#
[jgarber]
sup loqi
#
[KevinMarks]
But enclosure too
#
[KevinMarks]
There was some debate on gathering all attributes of a rel-linked url, but we limited it ones with a use case. We could review that list.
#
[jgarber]
I’ll draft up an issue on the microformats2-parsing repo later this evening with some example input and parser output. Not looking to reinvent the whole thing, but some feedback and extra definition in the spec would be helpful.
#
[KevinMarks]
So one use case is finding feeds - you check the list for rel=alternate, then look them up in rel-urls to provide the choice.
#
[jgarber]
Using the `type` key’s value in that example?
#
sknebel
[jgarber]: do the parsers do things not in the spec, or is the spec unclear?
#
[jgarber]
sknebel: Yes 😂
#
[jgarber]
Stepping out for some fresh air. Will follow up with an issue on the microformats2-parsing and post here when that’s up.
#
[KevinMarks]
Type and text/title yes