#microformats 2018-08-10

2018-08-10 UTC
GDiaX, tantek__, [jgmac1106] and [tantek] joined the channel
#
[tantek] Kevinmarks maybe ^^^? He’s got plenty of python and hero just experience afaik
#
[tantek] hero just = heroku
tantek, RoBz13, Ryuzaki, Adran9, CompanionCube29, hvxgr6, PaulCapestany, tx27, ascheel16, KartikPrabhu and barpthewire joined the channel
#
@swentel ↩️ @netsensei Aah, I can follow you now through the Microformats feed in my Indigenous client ;-) (twitter.com/_/status/1027832658481606656)
barbietoss, barpthewire, rud0lf24, L23529, [jgmac1106], [grantcodes] and [kevinmarks] joined the channel
#
[kevinmarks] I should give unmung.com a UX makeover
[jgmac1106], TehNut, [pfefferle], tantek, barpthewire, profall0, drot21, strugee and barpthewire1 joined the channel
#
ben_thatmustbeme Hmm make s ticket for every issue I find with the test set?
#
ben_thatmustbeme For every test that is. Several have implied name/url issues
#
sknebel nah, make a summary issue for similar issues. or submit a PR fixing it :D
barpthewire, tantek__, Sebastien11, [eddie], Humbedooh17, barpthewire1, vivus, planetmaker25, drakythe24 and Sitri17 joined the channel; vivus left the channel
#
ben_thatmustbeme hmm, you aren't supposed to imply name when a back-compat root right?
#
Zegnat nope
#
Zegnat At least, I seem to remember php-mf2 does not :P
#
Zegnat checks spec to be sure
#
ben_thatmustbeme i'm looking at the v1 tests and
#
ben_thatmustbeme https://github.com/microformats/tests/blob/master/tests/microformats-v1/hcard/single.html
#
ben_thatmustbeme https://github.com/microformats/tests/blob/master/tests/microformats-v1/hcard/single.json
#
Zegnat can’t seem to access microformats.org
#
Zegnat Uh oh: “Fatal error: Out of memory (allocated 786432) (tried to allocate 491520 bytes) in /var/www/html/wiki/includes/DefaultSettings.php on line 2598”
#
ben_thatmustbeme it was working a moment ago for me
#
aaronpk that poor server
#
ben_thatmustbeme still have the tab open, it specifically says (no backcompat roots)
#
KartikPrabhu yeah no implied rules for backcompat
#
KartikPrabhu of any kind
#
ben_thatmustbeme so that example above, the geo: should not have any properties
#
KartikPrabhu ben_thatmustbeme: in backcompat geo becomes "p-geo" so I think you should get "geo:[30.267991;-97.739568] "
#
aaronpk well now we know the cause of the server crashing https://media.aaronpk.com/file-20180810083653-3473.png
#
aaronpk a slight bump in traffic and it falls over
#
aaronpk "slight bump" meaning 2-3x the normal traffic
#
Zegnat Wow, we certainly been getting popular?!
#
Zegnat Maybe check that we don’t have another million fake users added? :(
#
aaronpk that's not an amount of traffic that should cause the server to fall over, so clearly apache needs some tuning
#
tantek__ we're not running anything complicated on Apache so that doesn't make any sense either
#
ben_thatmustbeme what version of php is it running?
#
aaronpk everything is oooold
#
aaronpk php 5.2
#
ben_thatmustbeme ick
#
ben_thatmustbeme can just up the memory limit in php.ini
#
ben_thatmustbeme at least see what the memory limit is
#
aaronpk it's at that unfortunate stage where nothing can be upgraded piecemeal anymore
#
aaronpk and upgrading php will probably break mediawiki because MW is too old too
#
aaronpk lol memory_limit => 32M
#
aaronpk something is eating memory tho https://media.aaronpk.com/file-20180810084826-7746.png
#
ben_thatmustbeme just looked at MW wiki version
#
ben_thatmustbeme yeah... not great
#
ben_thatmustbeme run top? sort by %mem?
#
aaronpk well it coincided with the increase apache traffic so it's almost certainly apache
#
ben_thatmustbeme oh yeah, fair point
#
aaronpk whoa it just ran out of memory again
#
aaronpk -bash: fork: Cannot allocate memory
#
ben_thatmustbeme wow
#
ben_thatmustbeme anything i can do to help?
#
ben_thatmustbeme also, KartikPrabhu: it should parse it as p-geo?? thats news to me
#
aaronpk any good links off the top of your head for optimizing apache/php settings?
#
ben_thatmustbeme in which case that geo: is very wrong
#
aaronpk this is a plenty powerful server, 6gb RAM, so we should be able to handle ridiculous amounts of traffic
d9b4bef910 joined the channel
#
ben_thatmustbeme damn, yeah
#
ben_thatmustbeme not offhand, no
#
ben_thatmustbeme MW 1.14 bug fixes: Fix excessive memory usage when parsing pages with lots of links
#
sknebel https://microformats.org.mirrors.sknebel.net/w/index.php?title=H-card#Backward_Compatibility - "geo - parse as p-geo h-geo including compat root class geo"
#
ben_thatmustbeme oh yeah, i thought you were saying just parse as p-geo
#
ben_thatmustbeme doesn't effect what the name: should be though does it?
#
aaronpk we need a microformats-meta channel heh
#
ben_thatmustbeme but that case would make a h-geo pointless
#
Zegnat I don’t know if it should be implying a name property on a backcompat h-geo ... I am going to guess no because it isn’t listed in the properties list? https://microformats.org.mirrors.sknebel.net/w/index.php?title=H-geo#Backward_Compatibility
Guest57736 joined the channel
#
ben_thatmustbeme aaronpk: you should know not to suggest channels around me :P
#
aaronpk notice i didn't make it clickable ;-)
#
ben_thatmustbeme doesn't matter, lol
#
ben_thatmustbeme aaronpk: a better question i guess would be how much it would take to upgrade MW
#
ben_thatmustbeme its only like 18 versions out of date lol
#
aaronpk well step 1 is inventory all the extensions used and see if they are available in the current MW version. then figure out whether the theme is compatible.
#
sknebel there should be a list of extensions on the wiki
#
aaronpk upgrading MW will almost certainly require upgrading PHP (which should happen anyway), but that means we need to also upgrade the wordpress that's there (which should also happen anyway)
#
sknebel and the question if preserving the theme is more important than having a functioning site
DarkMukke14 joined the channel
#
aaronpk right, we made the call on the indieweb wiki that it was not worth it
#
aaronpk er, that the theme was not worth it
#
sknebel (same with plugins, depending on the details)
#
sknebel only thing I've obviously noticed in the mirror is the <entry-title> thing ,but that doesn't mean it's the only one
#
aaronpk i know at least one of the extensions is obsolete thanks to it basically being rolled into core
#
sknebel https://microformats.org.mirrors.sknebel.net/w/index.php?title=Mediawiki-customization
#
aaronpk sknebel: is that a new MW version?
#
sknebel of course
#
sknebel not going to play software-archeologist and try to run outdated, possibly dangerously broken versions of software I barely understand on a modern system :D
#
ben_thatmustbeme only one extension installed, OverrideEntryTitle
#
ben_thatmustbeme its link to mediawiki.org is dead though
#
sknebel not done any customization to the mirror, just dumped the wiki contents in
#
aaronpk cool
#
aaronpk the best way to upgrade this is probably going to be to just set up a new one and run an export/import
#
aaronpk tho i can try using MW's upgrade process on a test instance first to see if it happens to work. i've never done one *this* old before
#
aaronpk watching top, i see an occasional apache process hit 100% cpu for a few seconds at a time
#
ben_thatmustbeme yeash
#
aaronpk i'm confused about the memory thing because it looks like there's 2.6gb free
#
aaronpk so i'm wondering if there's something causing a single apache process to eat up 2gb ram occasionally
#
aaronpk then it gets killed and the memory is freed
#
ben_thatmustbeme git a page with lots of links
#
ben_thatmustbeme hit*
#
ben_thatmustbeme not getting anything loading right now actually
#
aaronpk reading https://stackoverflow.com/questions/12015569/fatal-error-out-of-memory-but-i-do-have-plenty-of-memory-php
#
aaronpk been a while since i've used apache
#
ben_thatmustbeme well its out of memory again aaronpk
#
aaronpk i can't even use bash when it's out of memory
#
aaronpk tab-complete in bash even fails lol
#
ben_thatmustbeme what are the memory limits? cause that really shouldn't happen
#
aaronpk if i could use any commands i would tell you
#
aaronpk rebooting
#
aaronpk should give me a few minutes to poke around before it happens again
#
aaronpk just set MaxRequestsPerChild so that should keep the memory down
#
ben_thatmustbeme what version of apache?
#
aaronpk 2.2.14
iwaim joined the channel
#
ben_thatmustbeme https://httpd.apache.org/docs/2.2/mod/core.html#rlimitmem has had some success for others, stops apache from taking over everybit of memory it can
#
aaronpk hmm
#
ben_thatmustbeme suggested here https://stackoverflow.com/posts/12225357/revisions
#
aaronpk k we'll give that a shot too
jackjamieson joined the channel
#
ben_thatmustbeme well it seems to be running fine for the moment
tantek joined the channel
#
ben_thatmustbeme when we do update, can we get a working recaptcha?
#
aaronpk haha yes
#
aaronpk i think it might be possible to update that recaptcha plugin in place too tho
#
gRegorLove I wonder if we need the WordPress install going forward. We could move posts into wiki pages and restrict who can edit them.
#
aaronpk interesting idea
#
tantek__ updating the recaptcha plugin in place short of a massive upgrade is likely worth it
#
gRegorLove One less thing to keep up to date is my thinking.
#
tantek__ gRegorLove: that's good admin thinking
#
gRegorLove We don't use commenting, right?
#
tantek__ another consideration is, if we're going to switch from WordPress, can we switch to something that supports IndieWeb protocols?
#
tantek__ gRegorLove: we do, for a brief period after each blog post
#
aaronpk nothing is as mature as wordpress or MW, so it seems worse to install something else on that right now
#
tantek__ gRegorLove: maybe start documenting WordPress alternatives in a brainstorming section on /admin-to-do?
#
tantek__ has to run, will check later
barpthewire joined the channel
#
aaronpk ok now i'm completely stumped as to why the server is down
#
aaronpk 2.4gb free memory, loadavg is 0.25
Olipro4, iwaim_ and barpthewire joined the channel
#
aaronpk restarted apache with a lower default number of workers
iwaim_ joined the channel
#
aaronpk what the heck it just stopped again
#
aaronpk restarting apache makes it go again
strugee, arooni20 and [kevinmarks] joined the channel
#
sknebel ben_thatmustbeme: the name thing is interesting... I would have interpreted the parsing rule such that a name shouldn't be implied, but parsers seem to do it
iwaim_ joined the channel
#
gRegorLove In that instance, looks like php-mf2 isn't keeping track when geo gets updated to p-geo h-geo, so it thinks that's an explicit h-geo
#
gRegorLove I'll file an issue
#
sknebel mf2py does the same
#
sknebel I suspect with the same reasoning
Guest28044 joined the channel
#
Zegnat I do wonder though... don’t we *want* an implied name there?
#
Zegnat Or maybe not. The example isn’t able to make good use of it (that is: provide a non-numerical location name)
#
gRegorLove The spec refers to "only on explicit h-x" which I take to mean "explicitly authored"
#
gRegorLove With this example, if anything, "Brighton" makes more sense as a p-name
#
Zegnat Yes, I thought it might already be using that as a name... but it wasn’t. So ignore me.
#
sknebel this example is weird in that I don't think it defines a valid mf1 "geo" object either, since it sets no properties?
#
sknebel or is the combined for also acceptable?
#
gRegorLove Hm, good question
#
sknebel geo also isn't the only construct like this
#
sknebel h-card adr is another one
#
gRegorLove Yeah, /geo mentions this abbr usage
#
gRegorLove "If an <abbr> element is used for a property, then the title attribute of the <abbr> element is the value of the property, instead of the contents of the element, which instead provide a human presentable version of the value."
#
sknebel so it deos
#
gRegorLove Filed https://github.com/microformats/php-mf2/issues/195
#
Loqi [gRegorLove] #195 Fix backcompat parsing for geo property
#
Zegnat Does it matter what /geo says though? Does that carry over to mf2 parsers?
#
sknebel filed https://github.com/microformats/mf2py/issues/127
#
Loqi [sknebel] #127 implied, nested backcompat parsing does not use backcompat rules (e.g. hcard geo)
fryst24 joined the channel
#
gRegorLove I would expect the backcompat parsing to follow the <abbr title> parsing unless h-geo backcompat said otherwise
#
gRegorLove otherwise we'd get an h-geo in the parse results with no properties, no value
#
gRegorLove poor server out of memory again :/
iwaim, [tantek], iwaim_ and Loqi joined the channel
#
[tantek] No /geo does not carryover to mf2 parsing
#
[tantek] None of the old vocab specs should
#
[tantek] All back-compact parsing should be fully specific in mf2 vocab specs
#
[tantek] fully specified*
#
[tantek] abbr parsing is already specified in the mf2 spec
#
Zegnat Is it? For h-* objects?
#
Zegnat The spec says to parse class="geo" as a "p-geo h-geo", but the example in the test case does not contain any property elements at all
rogue2, gRegorLove and jackjamieson joined the channel
#
@taravancil ↩️ Yeah we’ve done a deep dive on ActivityPub, JSON-LD, microformats, and all the other attempts to standardize on that front, and we didn’t come out on the other end feeling confident in any one of them. (twitter.com/_/status/1028008463165804544)
cebor, [eddie], t0ne9, j12t, tantek__ and [kevinmarks] joined the channel
#
@kevinmarks ↩️ I'd love to talk to you again about microformats, but I know your json centric worldview makes that tricky. (twitter.com/_/status/1028026578637336578)
#
tantek__ ^^^ that makes me think we could maybe provide better docs (and a better name?) for mf2json
#
Zegnat http://microformats.org/wiki/microformats2-json ?
#
tantek__ so if folks wanted to they could just pass around mf2json, and thus be compatible with everyone else consuming mf2 via a JSON parser
#
tantek__ zegnat, yes, improving that page (and maybe name)
#
aaronpk yeah there's a lot of value in the microformats vocabulary even if you don't use it with HTML
iwaim joined the channel
#
Zegnat That page was written to explain the format of the JSON, not really as a usecase page. But there was a use at the time because more people started with micropub and microsub and were running into mf2 JSON
#
tantek__ yes that page needs lots of improvements in order to be a "first stop" reference
#
tantek__ simpler / better / more "real world" examples to illustrate each feature
#
tantek__ etc.
#
tantek__ like jf2 and jsonfeed, if mf2json had a better name, it may get more uptake
#
tantek__ I think one way to drive this would be from the Micropub perspective
#
tantek__ since Micropub normatively references mf2 for vocabulary, and then provides a JSON API
#
[kevinmarks] The Fritter/beaker people are very json centred
[iambismark] joined the channel
#
tantek__ kevinmarks, they are also some of the kindest most empathetic user-centric developers in the dweb "world"
#
tantek__ I met both Tara and Paul last week at dweb summit
#
tantek__ we found we (indieweb and beaker/dat) had *much* more in common
#
[kevinmarks] Yes, absolutely. I think they are great
#
tantek__ especially in contrast to many (most?) other dweb projects/protocols
[kiai] joined the channel
#
[kevinmarks] But they are not html fans
#
tantek__ they totally get UX-first, humans first
#
tantek__ that's not a fair characterization
#
tantek__ they don't dislike HTML
#
tantek__ it's just not their focus
#
[kevinmarks] No, but they think in json
#
tantek__ more like and they think in JSON
#
tantek__ I understand that aspect
#
[kevinmarks] The more we can work with them the better
#
tantek__ need to start a "brainstorming" section on http://microformats.org/wiki/microformats2-json, subsection "naming" to discuss better names
#
tantek__ kevinmarks indeed
#
tantek__ what if we put HTML and JSON examples side by side on each of the mf2 vocab specs?
#
Loqi http://meme.loqi.me/m/4vt_0_o_.jpg
#
tantek__ lol
#
aaronpk that would be very helpful
#
Zegnat Wasn’t someone working on that for the indieweb wiki? grantcodes perhaps..?
#
aaronpk oh yea the tab thing
jackjamieson joined the channel
#
[kevinmarks] My site has been on dat for quite a while
#
[kevinmarks] The "clone this site and edit it" thing in beaker is great, but the underlying protocol stuff is a lot more json flavoured
#
tantek__ no tab thing
#
tantek__ seeing them literally side by side is educational
#
[kevinmarks] Of all the distributed files people, they're the most indieweb
#
tantek__ hiding one while showing the other loses that
#
tantek__ it's hard(er) to build mf2 HTML examples that fit nicely in a narrow column, so you can put two columns side-by-side, but doable
#
[kevinmarks] You need a bit of mobile friendly css, but agreed
#
tantek__ (unlike pretty much any other attempt at HTML meta-markup
#
tantek__ )
indy joined the channel
#
[kevinmarks] You need to have a compact indentation model, as the extra [] we add give a lot of tabs in default json
#
tantek__ kevinmarks can you suggest docs/URL for a more compact JSON indentation model?
[dgold] joined the channel
#
@nagahitoyuki ブログに #Microformats のエラーが出てから #インデックス が三十件くらい減った。#アクセス も伸び悩むし、Googleでの #サイト評価 に #構造化データ の影響は大きいのか。updatedを記載しても表示しないと駄目なのかどうか、 #SearchConsole の結果から早く #何とかしたい。https://www.nagahitoyuki.com/ (twitter.com/_/status/1028042447228624896)
#
@nagahitoyuki ブログに #Microformats の #エラー が出てから #インデックス が三十件くらい減った。#アクセス も伸び悩むし、Googleでの #サイト評価 に #構造化データ の影響は大きいのか。updatedを記載しても表示しないと駄目なのかどうか、SearchConsoleの結果から早く #何とかしたい。https://www.nagahitoyuki.com/ (twitter.com/_/status/1028042946094874624)
#
@nagahitoyuki ブログに #Microformats の #エラー が出てから #インデックス が三十件くらい減った。アクセスも伸び悩むし、Googleでの #サイト評価 に #構造化データ の影響は大きいのか。updatedを記載しても表示しないと駄目なのかどうか、Search Consoleの結果から早く #何とかしたい。https://www.nagahitoyuki.com/ (twitter.com/_/status/1028043339034050561)
#
tantek__ wat
[cleverdevil], Dread3 and Gabemo28 joined the channel
#
tantek__ gRegorLove: re: newer hCalendar - sorta, in bits: http://microformats.org/wiki/hcalendar-brainstorming#how_to_use_with_HTML5
#
tantek__ and as evidence: http://microformats.org/wiki/hcalendar-parsing
#
gRegorLove Aha, missed that page.
#
tantek__ the need to do that (write *separate* parsing for hcalendar) was one of the nits that stuck in my head motivating subsequent brainstorming, design, iteration of microformats2 the next May
#
gRegorLove Interesting
#
tantek__ http://microformats.org/wiki/microformats2-origins
#
gRegorLove so hCalendar 1.0.1 and 1.1 were brainstormed but didn't make it to full specs?
#
tantek__ right, so they remained as errata / patches to 1.0
#
tantek__ there's probably some value in going through existing hCalendar implementations, seeing what got adopted from brainstorming etc., and formalizing those into a "frozen" version of those specs to document implementation interop
#
tantek__ and similar if people are publishing using those patterns, those can feed into back-compat parsing section in /h-event
#
tantek__ gRegorLove: that "frozen" version of hCalendar would be either 1.0.1 or 1.1 depending on the kind/number of changes accepted from brainstorming etc.
#
tantek__ would love to do a longer brainstorming session on that with you, since you've implemented a bunch of h-event processing code
#
tantek__ and publishing too vis-a-vis the indieweb /events page templates
#
gRegorLove Tangent: I need to check similar pages for hReview. I was looking for examples there with fn, photo, url properties *not* nested inside an `item`. (github comment: https://github.com/microformats/php-mf2/issues/155#issuecomment-377309540)
#
Loqi [gRegorLove] This is currently awaiting examples in the wild of hReview with `fn`, `photo`, or `url` properties that are _not_ nested in an `item`. I've been working through [hReview Examples in the wild](http://microformats.org/wiki/hreview-examples-in-wild#New_...
#
gRegorLove tantek__, Yeah, let me go through these other hReview pages more
#
gRegorLove oops, hCalendar-*
#
tantek__ cool - could definitely use another pair of eyes reviewing them for "hey this is obviously something that is/should be supported"
#
tantek__ or "hey this is nuts and not even close to workable"
#
tantek__ or somewhere in between :)
[kevinmarks] joined the channel
#
[kevinmarks] Indentation and whitespace is possibly an even bigger religious war than json vs html
[schmarty], barq13 and dh128 joined the channel
#
gRegorLove From hcalendar-brainstorming, I think we can add Tabular event calendars to the spec. hCard locations/attendees/organizers is already on the spec page as "may", so I think that's good.
#
gRegorLove Should add the <time> authoring to the spec