#dev 2016-07-14

2016-07-14 UTC
#
www.boffosocko.com
edited /events/2016-07-27-homebrew-website-club (+19) "/* Los Angeles */ Chris Aldrich RSVP yes"
(view diff)
#
www.boffosocko.com
edited /events/2016-07-27-homebrew-website-club (+93) "/* Where */ expections for Los Angeles area"
(view diff)
tantek, KevinMarks_, KevinMarks, gRegorLove, lukasros and cweiske joined the channel
#
pfefferle
are you there GWG?
#
Loqi
pfefferle: GWG left you a message on 7/11 at 6:47am: Not to be a pest, but any news on the PR?
#
pfefferle
good morning all
#
Loqi
rise and shine!
#
pfefferle
what about your german loqi?
#
cweiske
Loqi, verstehst du mich?
#
Loqi
is done
#
cweiske
warum bist du fertig Loqi ?
#
Loqi
grins profusely
#
cweiske
guten morgen Loqi
#
cweiske
are you playing with pokemons, Loqi ?
#
cweiske
yo Loqi
#
pfefferle
!tell GWG I also had problems with my crons... WordPress created a bunch of tasks and the process din't clean them correctly, so I used https://wordpress.org/plugins/advanced-cron-manager/ to clean up the schedule and now it works like a charm...
#
Loqi
Ok, I'll tell them that when I see them next
tantek joined the channel
#
miklb
aaronpk were you aware webmention.io was down?
KevinMarks_ joined the channel
#
j4y_funabashi
hey, anyone using flat-file storage - do you store extra hidden meta data with your files?
#
j4y_funabashi
I am storing my posts as html and I want to store my recieved mentions in the same way
#
cweiske
I have raw HTML files that get built into the final HTML files (adding navigation and framing)
#
cweiske
when I'll implement webmention display, I'll render the web mention files into the final HTML
#
j4y_funabashi
but mentions have extra metadata (source + target) that is not present in the post so I am wondering if I can store these variables in the html but dont know the best way to do that
#
cweiske
<meta name="webmention-source" content="..."/>
#
cweiske
same for target
#
cweiske
or, even better: <link rel="webmention-source" href="..."/>
#
j4y_funabashi
cweiske: yeah a rel link is a good idea, I want it to be parsed by mf2
#
j4y_funabashi
so when I accept and validate the mention I can inject a couple of rel links before saving the html
#
j4y_funabashi
are you planning to do something similar?
#
cweiske
stapibas, my webmention receiver, stores received mentions in mysql
#
cweiske
I've not decided yet if I'll dump them to the disk or not
#
cweiske
displaying mentions is not my top priority
#
j4y_funabashi
yeah fair enough - what are you working on at the moment?
#
cweiske
I finished my own search engine, search.cweiske.de and am now thinking about how to integrate my website screenshot service http://phancap.cweiske.de/phancap.phar/ as tool to archive linked pages
#
cweiske
either I integrate it directly into stapibas (which handles sending out linkbacks to linked pages, so the code is already there), or I'll make a standalone tool that subscribes to my hub for updates, and acts based on its notifications
#
cweiske
and then I need a way to make the archived pages usable/discoverable
#
cweiske
a lot to think about
#
petermolnar
j4y_funabashi I'm on-the-fly exporting wordpress to a plain-text-ish format, including comments
#
petermolnar
plain-text-ish as in as readable as possible, but the content is markdown
#
petermolnar
I've tried to migrate to grav and before this format I tried to export to a grav friendly YAML + MD file
#
petermolnar
but it didn't work out: grav lacks some features, especially in media management, which I'd need, like an imagemagick based engine
petermolnar joined the channel
#
petermolnar
so in the plain text file, I store meta as real text data
#
j4y_funabashi
petermolnar: that looks nice and neat, any reason you dont store as html + mf2?
#
petermolnar
simple: hard to read
#
petermolnar
this is a copy that on it's own can be parsed, read, and, in rare cases, printed
#
j4y_funabashi
cweiske: yeah archives are tricky
#
cweiske
petermolnar, do you have image galleries? how do you express them in markdown?
#
petermolnar
also, links become footnotes which is not shown in the example
#
petermolnar
I don't do image galleries any more, but I do have multiple images in blog posts; they are simle ![]() markdown images
#
j4y_funabashi
yeah I go back and forth on this one
#
j4y_funabashi
you could argue that html has all those properties, parse/read/print
#
cweiske
when I look at my blog posts, this could work, too - except for on-the-top-right floating images like in http://cweiske.de/tagebuch/noxon-gateway.htm
#
j4y_funabashi
I used to store everything as json but switched to html after reading http://notenoughneon.com/2016/5/29/purl-a-portable-content-store
#
cweiske
that's basically what I use now
KevinMarks joined the channel
#
petermolnar
cweiske ![](){} -> the {} is supported by makrdown extra specs and can contain .class or #id for css properties
#
cweiske
after reading this I know why I prefer rST over markdown
#
petermolnar
the reason why I decided to use plain text is because reading http://textfiles.com/magazines/LOD/ is still the same, but reading my own HTMLs from 16 years before is messy
#
petermolnar
cweiske I was considering rST, to be honest, I still am
#
petermolnar
althoug setext would probably be the best for long term :D
#
cweiske
too few structural features
#
cweiske
s/too few/not enough/
#
Loqi
cweiske meant to say: not enough structural features
#
petermolnar
I know, that's why it's immortal
#
cweiske
and that's why I cannot use it
#
cweiske
I require tables or definition lists
#
petermolnar
this is an interesting summary by the way: https://gist.github.com/dupuy/1855764
#
petermolnar
and there is also this: http://commonmark.org/
#
petermolnar
what is CommonMark?
#
Loqi
It looks like we don't have a page for "CommonMark" yet. Would you like to create it?
#
petermolnar
CommonMark is "A strongly defined, highly compatible specification of Markdown", in other words an attempt to make a common specification out of the countless Markdown versions out in the wild. See http://commonmark.org/
#
loqi.me
created /CommonMark (+252) "prompted by petermolnar and dfn added by petermolnar"
(view diff)
#
petermolnar
and there is always the option to export ot TeX ;)
#
cweiske
commonmark sounds like a good example for https://xkcd.com/927/
#
petermolnar
but I do appreciate it
#
tommorris.org
edited /site-deaths (+210) "/* Upcoming */ adding State"
(view diff)
#
voxpelli
for Micropub I save my posts with the e-* as markdown and the rets of the properties as YAML front-matter, so filed based storage, but not pure HTML
#
voxpelli
j4y_funabashi: cweiske: I wouldn't use <link rel="webmention-source" href="..."/> – if you're using a non-standard rel or mf2-property, then mark it up as such. See eg. http://microformats.org/wiki/microformats-2-faq#how_do_you_use_experimental_microformats_and_property_names
#
voxpelli
also don't see the need for a webmention-source – a h-entry or u-comment (u-comment's can't be standalone so h-entry is probably better) of the comment would contain that information already
#
voxpelli
creating new relation names or mf2 properties would according to how I perceive rfc6648 be considered fairly reckless without having gone done the research one could be assumed for a normally standardized property
#
j4y_funabashi
voxpelli: you are right, the source would be the entry's url property (if the author includes that)
#
voxpelli
j4y_funabashi: well, if the author doesn't include it, then you likely will, no? the h-entry you save will be your interpretation of the sent mention rather than the actual raw mention mf2, no?
#
j4y_funabashi
well, at the moment I am storing the source's html on the filesystem, then the mf2 json in my db cache
#
j4y_funabashi
and I want to be able to recreate the cache from the html
#
voxpelli
as long as you know the source URL you can always refetch the html
#
j4y_funabashi
so I am thinking the best way is to inject some metadata (the source and target) into the html before I save it
#
voxpelli
I'm saving the raw output of my mf2 parser alongside my interpretation of that data and the source URL and such in the database
#
j4y_funabashi
yeah I am trying to avoid extra http calls
#
j4y_funabashi
so if I have to recreate my dbcache I dont have to rely on the source of the mention still being around
#
voxpelli
I find YAML front-matter to be a pretty neat way of storing additional metadata in a file
#
voxpelli
keeps a clean separation between the html of the source post and your added metadata as well
#
cweiske
then you need Yet Another Parser
#
j4y_funabashi
although I think it would be nice to get all the info I need from the mf2 parser
#
voxpelli
j4y_funabashi: what language are you using?
#
voxpelli
a quick Google reveals quite a few fornt matter parsers for PHP, see eg: https://github.com/mnapoli/FrontYAML
#
cweiske
but you still need Yet Another Parser
#
voxpelli
cweiske: and what's the problem in that?
#
j4y_funabashi
as notenoughneon says in her post, it would be nice to parse your own posts the same way you parse any other url
#
voxpelli
some languages like Python even have YAML parsers built in as far as I know
#
cweiske
more code -> more problems
#
voxpelli
cweiske: premature optimization is the root of all evil...
#
cweiske
Lets Use 10 Different Formats Now. We Can Reduce That Later.
#
j4y_funabashi
voxpelli: thanks for the microformats link, will have a read through
rMdes joined the channel
#
j4y_funabashi
are custom rel values a microformats thing?
#
voxpelli
j4y_funabashi: if you would like to follow cweiske's advise and at all costs avoid extra dependencies in your code, then you can always do JSON Front Matter instead of YAML: https://gist.github.com/voxpelli/a4fbc66e9bc6dbfec8fa9b7654fdee59 Less common, harder to read, but less code
#
voxpelli
j4y_funabashi: I think http://microformats.org/wiki/rel-registry is the registry for rels, athough then there also are https://tools.ietf.org/html/rfc5988
#
voxpelli
j4y_funabashi: custom rels according to at least rfc5988 should be defined as URL:s – such as rel="http://voxpelli.com/rel/my-personal-rel"
#
j4y_funabashi
for me it isnt really about avoiding dependencies I would just rather not write/read another file if I dont have to
#
voxpelli
j4y_funabashi: it wouldn't be another file, it would be another part of the same file
#
cweiske
.. still another parser
#
j4y_funabashi
ah ok, I see what you mean
#
voxpelli
cweiske: JSON is part of most languages core libraries and as such won't change, and won't be removed
#
cweiske
I consider extracting some JSON from some HTML comment as parsing
#
voxpelli
Yes, you have said that you don't like adding code or parsers, we get it
#
voxpelli
So it's better for j4y_funabashi to mix his own data with the data downloaded from another site?
#
voxpelli
I think it would be a good idea to separate the metadata, which was the original question how to add, from the data downloaded from another site and a common technique for that is front matter
#
j4y_funabashi
at first I thought I could put the source+target in the filename
#
j4y_funabashi
but I prefer the idea of having those variables inside the file
#
voxpelli
using a standard serializer, be it JSON, YAML or HTML would probably be easiest/safest
#
voxpelli
partly because the data that you are going to save comes from a client, and you should never trust the client, but expect them to do the craziest things
#
j4y_funabashi
yeah I dont ever display the mention html, I render the mf2 version
#
j4y_funabashi
I think I am going to go with injecting 2 rel links to the head of the document for now
#
j4y_funabashi
and prefix them so they dont interfere with anything else
#
j4y_funabashi
aruna-webmention-source + aruna-webmention-target
#
voxpelli
j4y_funabashi: if it's rels then the more common way is to go with a url then
#
voxpelli
rel="http://j4y-something.com/webmention/source"
#
voxpelli
j4y_funabashi: any thoughts on how you will handle it if those are already tehre?
#
voxpelli
you probably need to ensure that none of your metadata tags has been added by the source page (or else you are trusting the client ;) )
#
j4y_funabashi
heh good point
#
j4y_funabashi
but yeah I can remove any rels with those values before injecting my own
#
j4y_funabashi
to avoid that kind of attack
#
GWG
Morning
#
Loqi
GWG: pfefferle left you a message 5 hours, 18 minutes ago: I also had problems with my crons... WordPress created a bunch of tasks and the process din't clean them correctly, so I used https://wordpress.org/plugins/advanced-cron-manager/ to clean up the schedule and now it works like a charm...
#
GWG
Will have to try that
#
GWG
pfefferle, thanks for merging that quick change. I am going to experiment with logging.
#
pfefferle
GWG good morning
#
GWG
pfefferle, any thoughts on logging webmentions to post meta?
tantek joined the channel
#
miklb
!tell aaronpk were you aware that webmention.io was down with a 502
#
Loqi
Ok, I'll tell them that when I see them next
#
tantek
j4y_funabashi: for another flat file storage example, I've documented some of Falcon's storage approach here: https://indieweb.org/Falcon#Storage_format
#
aaronpk
Crap. Wil fix soon
#
Loqi
aaronpk: miklb left you a message 12 minutes ago: were you aware that webmention.io was down with a 502
#
tantek
petermolnar: would be interested in your opinion (in terms of markdown replacements) with http://tantek.com/w/Markdown
#
petermolnar
I'll check it
#
cweiske
we should document 502 errors as criticism to nginx
#
Loqi
it is probable
#
sknebel
wrong answer, Loqi
#
tantek
what is a 502?
#
Loqi
502 is an HTTP/1.1 Status Code returned from a webserver to signify that the Proxy Gateway for the requested URI was not found https://indieweb.org/502
#
tantek
what is a Proxy Gateway?
#
Loqi
It looks like we don't have a page for "Proxy Gateway" yet. Would you like to create it?
#
aaronpk
it's not nginx's fault
#
aaronpk
i never have this problem with PHP
#
aaronpk
PHP behind nginx
#
aaronpk
sometimes when I reload the nginx config, passenger forgets about the ruby app
#
petermolnar
I did have it with nginx + php when the php-fpm service went down
#
petermolnar
but I don't see why this is criticism for nginx
#
petermolnar
it's a totally valid error
#
aaronpk
what caused the php-fpm service to go down? I've never had that happen yet
#
petermolnar
aaronpk self-compiled php7-fpm :)
#
petermolnar
I hit some random errors with it
#
petermolnar
but to be honest, it's extremely rare
#
cweiske
petermolnar, if php's fatal memory error is valid critic to wordpress, then 502 is valid critic for nginx - see https://indieweb.org/WordPress#Fatal_Error_memory_exhausted
#
petermolnar
that's not a valid critic either
lukasros joined the channel
#
aaronpk
IIRC that got moved to the PHP page
#
cweiske
it's still on wordpress
#
cweiske
with tantek noting that wordpress should catch it
#
petermolnar
and how exactly an oom killer-type error could be caught from within a script?
#
petermolnar
it's not a php exception, it's an instant die
#
cweiske
I didn't get an answer to this
#
miklb
iirc I got those php-fpm errors when updating Debian and a cache directory getting lost and not being writable
#
cweiske
tantek and frankly if WP *can't* be fixed, then that's a criticism too - that this is a non-fixable problem with WP
#
sknebel
(aaronpk: doesn't Loqi quote chat-archive links anymore?)
#
aaronpk
oh! huh
#
aaronpk
ah I didn't enable it for this channel
#
begriffs.com
edited /silo-quits (+392) "/* Silo quitters 2016 */"
(view diff)
gRegorLove joined the channel
#
petermolnar
!tell tantek re: http://tantek.pbworks.com/w/page/59905776/Markdown - most of the ideas are valid; I like the multi-line image w/ optional url and alt-ext and the bold vs italics, but I would miss the reference-style url options, which I do use now
#
Loqi
Ok, I'll tell him that when I see him next
#
petermolnar
!tell tantek maybe forking/extending https://github.com/erusev/parsedown-extra would be a good idea to support this type of subset
#
Loqi
Ok, I'll tell him that when I see him next
petermolnar joined the channel
#
petermolnar
!tell tantek this is also a good read on a similar topic: https://gist.github.com/dupuy/1855764
#
Loqi
Ok, I'll tell him that when I see him next
#
KevinMarks
Cwieske can you not put the source in as a base url ? That should fix relative url parsing too.
#
KevinMarks
The reason I like commonmark is that is based on research. Also http://johnmacfarlane.net/babelmark2/ is very good
KevinMarks and KevinMarks_ joined the channel
#
gregorlove.com
edited /CommonMark (+28) "link dfn, see also"
(view diff)
KevinMarks joined the channel
#
www.orbific.com
edited /2016/Brighton (+51) "/* Participating */"
(view diff)
#
GWG
Howdy
#
GWG
Hello, voxpelli. I am working on the issue you brought to my attention
#
voxpelli
Nice :)
#
GWG
pfefferle accepted a PR to move an existing hook that only triggered on send success to trigger on all sends
#
GWG
So I can write a logging function to trigger off of it, either built-in or separate
#
@voxpelli
Celebrating the successful large update of https://webmention.herokuapp.com/ yesterday by releasing a couple of new accounts for people to sign up on
(twitter.com/_/status/753670585067114497)
#
GWG
For now, I may just borrow petermolnar's
#
miklb
voxpelli I meant to tell you I switched back to wm.io so I could use the updated plugin for Jekyll and have cached copies of webmentions. Some day may look at adapting the plugin to use yours though.
#
miklb
I do need to look into why I have the duplicate mentions
#
voxpelli
miklb: I think an issue I filed in Bridgy was solved – could have affected Webmention.io as well
#
GWG
dedupe function needed?
#
voxpelli
GWG: yeah, proper u-url based dedupe function would have prevented the issue Bridgy had from carrying through all the way
#
voxpelli
miklb: if you want to build a plugin some day then you can look at what Stuart built on top of my endpoint: http://www.kryogenix.org/days/2014/11/29/enabling-webmentions/
#
Loqi
Enabling Webmentions
#
voxpelli
miklb: one could imagine a bridge to Micropub as well so that comments were imported using Micropub
#
miklb
ooh. That would be neat
#
voxpelli
as my endpoint has a realtime API one could even make it so that those Micropub requests happens in realtime :)
#
voxpelli
I'm not going to focus on that though as for Jekyll the small delay in building will cause problems when doing Salmentions
#
miklb
first on my todo list is a redesign/re-structure now that I have micropub
#
voxpelli
but one could build it as a standalone service :)
#
voxpelli
has even considered getting Pubsubhubbub going so that he has a more standard realtime API
#
miklb
needs to do some debugging for the bookmark type. When syndicating with bridgy, it only seemed to use the title and not content for the tweet
tantek joined the channel
#
loqi.me
created /content-addressable_storage (+82) "prompted by gRegorLove and dfn added by gRegorLove"
(view diff)
#
loqi.me
created /your_interest (+19) "prompted by KevinMarks and dfn added by gRegorLove"
(view diff)
#
thraxil.org
edited /2016/Brighton (+20) "/* Participating */"
(view diff)
#
thraxil.org
created /User:Thraxil.org (+17) "Created page with "is Anders Pearson""
(view diff)
#
thraxil.org
edited /2016/Brighton (-6) "/* Participating */"
(view diff)
#
thraxil.org
edited /2016/Brighton (+38) "/* Participating */"
(view diff)
#
rossadavis.com
created /Template:ross (+147) "Created page with "<span class="h-card" style="white-space:nowrap">{{sparkline|https://rossadavis.com/images/BigBio2.png}} [[User:rossadavis.com|Ross A Davis]]</span>""
(view diff)
tantek joined the channel
#
tantek.com
edited /why (+179) "/* Site loss */ Google removes artist’s blog and a decade of his work along with it (and their email too)"
(view diff)
#
tantek.com
edited /lulz (+69) "just setting up my statuses"
(view diff)
#
www.boffosocko.com
edited /Homebrew_Website_Club (+7) "kickstarting Los Angeles"
(view diff)
#
tantek.com
edited /marked_safe (+123) "/* Facebook */ screenshot of notification marked safe"
(view diff)
#
tantek.com
edited /marked_safe (+11) "/* Facebook */ date"
(view diff)
#
tantek.com
edited /safety_check (+86) "add screenshot date first"
(view diff)
#
tantek
apparently FB also prompts people who have their current city set
#
Loqi
tantek: petermolnar left you a message 6 hours, 41 minutes ago: re: http://tantek.pbworks.com/w/page/59905776/Markdown - most of the ideas are valid; I like the multi-line image w/ optional url and alt-ext and the bold vs italics, but I would miss the reference-style url options, which I do use now
#
Loqi
tantek: petermolnar left you a message 6 hours, 40 minutes ago: maybe forking/extending https://github.com/erusev/parsedown-extra would be a good idea to support this type of subset
#
Loqi
tantek: petermolnar left you a message 6 hours, 39 minutes ago: this is also a good read on a similar topic: https://gist.github.com/dupuy/1855764
#
tantek
and you can ask a friend if they are safe also https://indieweb.org/safety_check
#
kevinmarks.com
edited /content_addressing (+81) "/* See Also */"
(view diff)
#
tantek.com
edited /events/2016-07-27-homebrew-website-club (+152) "/* RSVP */ link LA indie event and FB POSSE copy!"
(view diff)