#dev 2016-09-01

2016-09-01 UTC
doesntgolf joined the channel
#
tantek
just saw the discussion of IndieAuth
#
tantek
aaronpk: where are you seeing "some people to not take it seriously as a profile of OAuth 2"?
KevinMarks joined the channel
#
miklb
hmm. does Loqi not parse links in -dev?
#
miklb
The Readability bookmarking service will shut down on September 30, 2016.
#
aaronpk
no mf2 on medium posts :-(
tantek, KevinMarks, miklb and cweiske joined the channel
#
cweiske
there is still a bug; "noindex" pages are shown in the results - e.g. join messages: http://indiechat.search.cweiske.de/?q=%22joined+the+channel%22
#
cweiske
but it works.
#
cweiske
now I can finally find when other people mention me: http://indiechat.search.cweiske.de/index.php?q=cweiske+-nick%3Acweiske
KevinMarks_ joined the channel
#
cweiske
aaronpk, I get connection timeouts on my server from chat.indieweb.org
#
cweiske
do you have some kind of rate limiting?
cmal, loicm_ and loicm__ joined the channel
#
przeorski.pl
edited /ReactJS (-14) "reactjs website update"
(view diff)
loicm__, pfefferle, AngeloGladding and doesntgolf joined the channel
#
GWG
pfefferle is here? I have been wanting to talk to him.
#
aaronpk
cweiske: no rate limiting, might be hitting max nginx workers or something
#
aaronpk
once I switch to disk based logs instead of DB I'll take a look at it
#
cweiske
to increase initial import speed i'll extend phinde so I can simply add a list of URLs that should be indexed, and will extend chat.indieweb.org to dump a list of *all* urls
#
cweiske
based on the db
#
cweiske
this will remove all the crawl time
#
aaronpk
all URLs?? I'm not sure that page will ever finish generating
#
aaronpk
is crawling the day permalinks still too slow?
#
cweiske
1s to extract the links and check if they are already in ES
#
cweiske
5s to extract and add to ES
#
cweiske
sounds like a performance problem with my ES instance
#
cweiske
but you're right, i'll probably trying to optimize the wrong issue
#
loqi.me
edited /UGT (+70) "/* See Also */ new section"
(view diff)
#
aaronpk
What is UGT?
#
Loqi
Universal Greeting Time is the convention of greeting people in online chat with "good morning" regardless of the local time of day of either person https://indieweb.org/UGT
KevinMarks and cweiske joined the channel
#
ben_thatmustbeme
don't know everything thats possible yet, but what are people interested in as far as mobilepub v3 functionality?
#
ben_thatmustbeme
i'm thinking i'm going with a more single view layout for post editing, just add and remove fields
#
ben_thatmustbeme
also ability to recall and edit saved offline posts
loicm__, tantek, cmal and gRegorLove joined the channel
#
gregorlove.com
edited /Readability (+92) "link dfn, note site-death date, see also"
(view diff)
#
gregorlove.com
edited /site-deaths (+704) "/* Upcoming */ Readability"
(view diff)
#
gregorlove.com
edited /site-deaths (+20) "archive Togethera"
(view diff)
#
gregorlove.com
edited /site-deaths (+36) "archive Google Drive web hosting, bump down "in months""
(view diff)
cweiske joined the channel
#
gRegorLove
This is apparently going again in 11 days and I don't see any notice of it on the page. Yikes. https://pip.verisignlabs.com/
#
gRegorLove
s/again/away/
#
cweiske
I known now why crawling is so slow. Before adding a new URL to elastic search, I'm checking if the URL is already know
#
cweiske
that "exists" query takes 40ms
#
cweiske
for a day page with 100+ log line URLs this takes too long
#
plindner
I think it’s both..
#
plindner
nm: wrong channel
#
KevinMarks
Can you use memcache for that, or are the urls sparse enough that it doesn't help?
#
cweiske
I have no idea why ES is so slow
#
cweiske
I'm very green regarding ES
#
aaronpk
cweiske: would it be faster to not check if it exists and just overwrite it if it does?
#
aaronpk
depends on how long a write takes i guess
KevinMarks and tantek joined the channel
#
cweiske
aaronpk, they might be already indexed, and I don't want to throw all that info away
#
cweiske
then I would have to re-index
#
aaronpk
what info?
#
aaronpk
for my site search, I just replace the ES record every time I update the post, whether or not the record exists before
#
cweiske
aaronpk, I have different processes for indexing and crawling
#
cweiske
at first the URL is added to the index when it's detected, marked as to-crawl and to-index
#
cweiske
the crawl process then extracts all urls from it and marks it as crawl-done
#
cweiske
and the index process extracts the relevant content from the URL and updates the document with that info
#
aaronpk
so you're using ES as a kind of queue?
#
cweiske
no, I'm using gearman as queue
#
cweiske
but keep info about the current state in ES
#
aaronpk
if you move your "to-index" list outside of ES then you can always replace the ES record with the new doc and not bother checking if it exists
#
cweiske
and I check if the document exists because I don't want to re-crawl and re-index it
#
cweiske
it's an existance check. I don't understand how it can be that expensive
#
cweiske
but I got an answer in the forum already
#
cweiske
I'll check all URLs at once with a single query; this is hopefully faster
#
aaronpk
so querying really just does take that long no matter what?
#
cweiske
yes. 40ms for a simple "does a document with URL $x exist"
kants joined the channel
#
aaronpk
i guess it's an HTTP request so you also have that overhead
#
KevinMarks
I may just edit the php so that it returns 'article' instead of 'note' as the databases are so tangled
#
tantek
KevinMarks, sounds like there may need to be some rethinking how WordPress works with modern varieties of posts
#
tantek
especially with respect to import/export
#
KevinMarks
I think this is Post Kinds making poor assumptions.
#
KevinMarks
'note' default doesn't really make sense in wordpress, as the ux default is article.
#
tantek
perhaps the ux default of "article" doesn't really make sense, as social media UX has moved on from that
#
tantek
article as default UX is an obsolete left-over from mid-2000s blogging UIs
#
KevinMarks
As opposed to early 2000s, where note was the default
#
ben_thatmustbeme
ohh, fun, mobilepub v3 will be able to recall offline posts, edit, and resave
loicm__ joined the channel
#
tantek
sweet ben_thatmustbeme
#
ben_thatmustbeme
I was a bit hesitant going to another JS framework, but this is working out REALLY nice
#
ben_thatmustbeme
will even use routing correctly, so back arrow navigation and such will be fine
#
bret
ben_thatmustbeme: what did you use?
#
ben_thatmustbeme
bret: aurelia.io
#
ben_thatmustbeme
really liking it
#
aaronparecki.com
edited /Special:Log/block () "blocked [[User:Przeorski.pl]] with an expiry time of infinite (autoblock disabled): Spamming links to external sites: only edits were adding an external link https://indieweb.org/Special:Contributions/Przeorski.pl"
(view diff)
loicm__ joined the channel
#
tantek
is that our first?
#
aaronpk
congrats?
#
tantek
and .pl (Poland?) is not a cheap ccTLD
#
tantek
well Pzeorski, hope it was worth it
#
bret
ben_thatmustbeme: are those attributes used in templates specific to aurelia?
#
bret
or is it some kind of web component thing
#
bret
I've been mucking about with https://github.com/yoshuawuyts/choo looks similar
loicm__ joined the channel
tantek and doesntgolf joined the channel
#
tantek.com
edited /ReactJS (+129) "make definition more readable and explain why it relates to indieweb, make criticism specific"
(view diff)
#
tantek.com
edited /ReactJS (+44) "dfn, also ReactJS"
(view diff)
#
tantek.com
deleted /React "Deleted to make way for move"
#
tantek.com
moved /ReactJS to /React "use canonical name"
#
tantek.com
edited /framework (+211) "attempt a more direct (less abstract) definition, reduce introduction of new jargon, example React, contrast with library"
(view diff)
#
tantek.com
created /library (+617) "stub with dfn, software vs physical, examples, see also"
(view diff)
#
tantek.com
edited /cassis.js (-14) "technically not a framework, since your code is still JS and/or PHP"
(view diff)
loicm__ joined the channel
#
tantek
so, frameworks
#
tantek
I think since people primarily use it as a framework (i.e. as a replacement for HTML, CSS, DOM), it's a framework
#
tantek
aside: Quora js;dr :P
#
sknebel
it's a tool to manage the DOM, and with similar logic many template languages are a framework as well?
#
tantek
sknebel examples?
#
sknebel
for what?
#
tantek
do you mean like Mustache is a framework replacement for HTML?
#
tantek
"many template languages"
#
ben_thatmustbeme
Bret what attributes?
#
sknebel
I'm not sure how mustache works internally, but generally all that do not only work on string-level but create individual dom-nodes from an internal representation mostly do "here is a template, here are some values, please fill them in and insert into the DOM here"
#
sknebel
React does the same, with the addition of logic to update them more efficiently if you change values
#
sknebel
but it's always my code calling react with "here is an update, please make the DOM match that"
#
sknebel
(from what I remember, I've only used it for a university project last year)
#
bret
ben_thatmustbeme: the for attributes that let you do for loops in your template
#
ben_thatmustbeme
Ahh no that's Aurelia
#
sknebel
looking up mustache it works on strings, so it is a different approach (tantek)