@mdom do you create 40mb traffic with 304 Not modified responses or do you fetch the entire resource?

I think twtxt scales pretty decent by using HTTP Features like Cache-Control. #scalability #twtxt

There is a nice paper on PULL or PUSH when it comes to scaling a justin bieber twitter account. http://research.yahoo.com/pub/3203 In this special case, twtxt's design is actually good! #scalability #twtxt

@mdom I would like to implement server and client, too ;). Since it somehow fixes the nervous polling, which https://web.twtxt.org does.

@kas will you whitelist twtxt-ui/registry in your robots.txt again?

@kas I updated the registry to interpret the robots.txt (your entries won't be indexed anymore) + the Crawl-Delay directive of robots.txt https://en.wikipedia.org/wiki/Robots_exclusion_standard#Crawl-delay_directive

@kas I changed it to 100 times a day with https://github.com/DracoBlue/twtxt-registry/commit/8cb58f8a37ad51bbfb56bd6e7235bfd092aae0ba ? Is that ok for you?

@kas Is fetching every 10 minutes for the registry and the webclients ok? Maybe we need something like "/changefreq hourly" as meta data for registries/clients.

@kas btw. 37.120.16.189 is not my ip ;)

@kas ok, I disabled it. will work on a fix tomorrow. sorry for the issues. :(

@kas wtf. what's wrong? I will remove them now.

About referencing tweets: how about md5(TWTURL\tTIMESTAMP\ttxt) as "id" of the tweet?

@mdom I meant the websocket part

@kas ok :(. Will do when back at PC. What is the problem?

From now on https://web.twtxt.org will only deliver the latest 20 twtxt messages (/user) to save bandwith. 304 Not Modified via E-Tag is also supported.

@mdom sounds great. I would really like to add it to registry.twtxt.org, too! Can you create a PR for https://github.com/buckket/twtxt/blob/master/docs/user/registry.rst and add documentation for this?

@mdom looks nice!

/follow twet https://lublin.se/twet.txt

/follow dracoblue https://dracoblue.net/twtxt.txt

I finished the next alpha of https://web.twtxt.org/. It has now "github" login (files at username.github.io/twtxt.txt) and stores all /follow+/unfollow things in the twtxt.txt.

/follow adiabetic https://www.frogorbits.com/twtxt.txt

/follow xena https://xena.greedo.xeserv.us/files/xena.txt

/follow twtxt https://buckket.org/twtxt_news.txt

/follow reednj http://reednj.com/reednj.twtxt.txt

/follow quite https://lublin.se/twtxt.txt

/follow kdave https://kdave.github.io/twtxt.txt

/follow kas https://enotty.dk/twtxt.txt

/follow horsemans http://horsemans.co/horsemans.txt

/follow erlehmann http://daten.dieweltistgarnichtso.net/tmp/docs/twtxt

/follow david http://post.aldebaran.uberspace.de/twtxt/david.txt

/follow buckket https://buckket.org/twtxt.txt

/follow mdom https://mdom.github.io/twtxt.txt

/twturl https://dracoblue.net/twtxt.txt

@reednj manually ;). I thought about it, but since the usernames can be different I droped that idea.

@mdom do you really need a second config file? Can't you use an own [txtnix] section in twtxt's config file? http://twtxt.readthedocs.org/en/latest/user/configuration.html#configuration

@reednj_tw yep. Etag + If-None-Match is way better, because it usually works with md5 on the content. https://en.wikipedia.org/wiki/HTTP_ETag /cc @buckket

@kdave do you want a subdomain of twtxt.org?

@kdave I can see if I can create a version, which tries to store the info, and falls back to cookies or no storage at all.

@kdave does the localstorage version not work in your fav. browser?

@kdave ah .. missread your message. What alternative way would you use for the web version? I would like to store the info within a config file, too. E.g. the twtxt.txt itself as meta data.

@kdave https://web.twtxt.org/ works with localstorage already in the browser. (for following users)

@kdave I don't think, it's necessary to merge http+https. People can add themself only with https. /cc @david

/me added /me support to https://web.twtxt.org

@mdom I am working on the same thing for https://web.twtxt.org right now, too! :)

@buckket yep ;). I think the ideas in #48 are already quite good!

/me is testing the proposed format of meta data in https://github.com/buckket/twtxt/issues/48

/twturl https://dracoblue.net/twtxt.txt

/nick dracoblue

@reednj the .txt files look good. Can you provide the mentions also by url? Ideally with /api/plain/mentions like described at http://twtxt.readthedocs.org/en/latest/user/registry.html#query-for-mentions

@reednj please upgrade your client, to have proper mentions in your tweets ;)

yep! nice ;)

@kas thank you!

Created nice API-Doc with "Try it out!"-forms for the registry api at https://registry.twtxt.org/swagger-ui/

@kas I added memcached and user agent to my registry, so you should see some 304 now from me!

@xena @twtxtlist I guess there is something wrong with the time at your side. It tweets as '2016-02-13T03:28:46+00:00' but, it's still 2016/12/02 http://www.timeanddate.com/worldclock/timezone/utc

@xena happens ;)

You have to use javascript:fetcher.login('https://example.org/twtxt') to login. I will add UI for that later.

added mentions-support by querying registry.twtxt.org to https://web.twtxt.org/

@kas that's the server for https://web.twtxt.org - will add proper User-Agent-String + Modified+Since soon. Fetches every minute if users are using it. Load is ok? /cc @quite

@buckket cool

@kas I implemented a proposal for such api here https://github.com/DracoBlue/twtxt-registry. Example response is: https://registry.twtxt.org/api/plain/users

Hey @reednj and @xena! do you want to work on an effort to streamline #twtxt registry apis? My proposal: https://github.com/DracoBlue/twtxt-registry

working on a web-ui for twtxt (preview at https://web.twtxt.org/ )

@buckket something like https://gitter.im/ or a freenode channel?

@buckket looks nice ;)

RT @icecrime: We're happy to announce the birth of @Docker Engine 1.10. It weighs 358797 LoC, and makes its 209 parents proud. https://t.co…

RT @lhtdesignde: having some fun with dashing.io and #raspberry hacking. great framework. #dashingio https://t.co/Bszu8V4mQw

RT @brucel: