@prologic@twtxt.net sounds about right. I tend to try to build my own before pulling in libs. learn more that way. I was looking at using it as a way to build my twt mirroring idea. and testing the lex parser with a wide ranging corpus to find edge cases. (the pgp signed feeds for one)

@prologic@twtxt.net the add function just scans recursivley everything.. but the idea is to just add and any new mentions then have a cron to update all known feeds

@prologic@twtxt.net yeah it reads a seed file. I’m using mine. it scans for any mention links and then scans them recursively. it reads from http/s or gopher. i don’t have much of a db yet.. it just writes to disk the feed and checks modified dates.. but I will add a db that has hashs/mentions/subjects and such.

@prologic@twtxt.netd It is pretty basic, and depends on some local changes i am still working out on my branch.. https://gist.github.com/JonLundy/dc19028ec81eb4ad6af74c50255e7cee

@lyse@lyse.isobeef.org @prologic@twtxt.net very curious… i worked on a very similar track. i built a spider that will trace off any follows = comments and mentions from other users and came up with:

twters:  744
total:  52073

curl https://raw.githubusercontent.com/jointwt/we-are-twtxt/master/we-are-twtxt.txt | grep -v '^niplav ' | field 2 | xargs curl ^/dev/null | grep niplav here we go

@prologic@twtxt.net “What the hell does cartesianly mean?! 😳” Me inventing a new adjective for “there is only a one-way flow of information” ^^

Maybe I could write a script that trawls the we-are-twtxt list and notifies me when I’m mentioned–the name should be unique enough.

Disadvantage of twtxt: less incentives to reply to people since it’s not certain they’ll ever see it. My current solution to that is to follow everybody on the we-are-twtxt and only unfollow if they twt a lot of stuff I’m not interested in

@felixp7@twtxt.net “Yo, crypto-heads. Encrypted communication doesn’t protect your privacy. Laws …” I guess crypto-heads are often happy with acknowledging this, but also arguing that e.g. deniable crypto is a useful way out here (and, of course, just doing illegal stuff if it gets really rough).

FOLLOW: @xuu@txt.sour.is from @niplav@niplav.github.io using txtnish/0.2

Following with interest the developer activity regarding the twtxt spec at twtxt.net

@prologic@twtxt.net I am trying to cut costs, so I deleted all my Digital Ocean droplets. For a month I will be using a free s390x VPS, so I needed to adapt some of my Docker images, scripts and configurations for that architecture. Also took another chance with Traefik Proxy, I ditched it long ago for nginx-proxy, but this time I made it work. #h5nn5tq

@prologic@twtxt.net that I do. lol. I am xuu on hackint.org and freenode

@prologic@twtxt.net sure. I don’t use signal much because I have to disclose my personal phone. Telegram? https://www.t.me/xypheri

@xuu@txt.sour.is @prologic@twtxt.net Your feed was great for catching edge cases ;)

@prologic@twtxt.net https://github.com/JonLundy/twtxt/tree/xuu/integrate-lextwt I made a stats command for the new parser that extracts a bunch of info about a twtxt file. run like: go run ./cmd/stats https://twtxt.net/user/prologic/twtxt.txt

FOLLOW: @xuu@txt.sour.is from @lyse@lyse.isobeef.org using twtxt/1.2.3

@prologic@twtxt.net yep!some of the lexer is directly copied from monkey-lang. love that book series.

@prologic@twtxt.net ah I need to add an edge case for naked urls with fragments.

@prologic@twtxt.net yep. it actually extracts everything at parse time. like mentions/tags/links/media. so they can be accessed and manipulated without additional parsing. it can then be output as MarkDown

@adi@twtxt.net @prologic@twtxt.net using regex. which can be a rather inexact science ;)

@prologic@twtxt.net kinda.. It can parse the twts into an AST.. but most of the formatting out expects a string to do regex over rather then the parsed AST. thats what i am working out next.

@prologic@twtxt.net as promised! https://github.com/JonLundy/twtxt/blob/xuu/integrate-lextwt/types/lextwt/lextwt_test.go#

the lexer is nearing completion.. the tough part left is rooting out all the formatting code.

the !zet I have written for my wiki is probably going to end up replacing what I’ve been using twtxt for these past few months. this means less spam here. you’re welcome.

@prologic@twtxt.netdd ooh I am adding that to my test suite

@prologic@twtxt.net @gareppa Tis fake.. that is the name of the tower in Die Hard. A movie that takes place on Christmas Eve. The actual name of the Nakatomi Plaza is the FOX Plaza.

@prologic@twtxt.net I agree, should have asked for help. Neither the bridge network nor any other bridge-based custom network allowed the containers to contact external servers, it might have been an iptables issue. #twflvza

@prologic@twtxt.net went over to watch this one in my home town last night. https://youtu.be/kUZB0_Jx3iE

#ChristmasLights #2020

@oevl@twtxt.net @prologic@twtxt.net (#) for the most part a subject is just the content in the perens. Usually it’s a tag. It appears near the start after any mentions. It can also contain text like (re: subjects)

WEBMENTION: @xuu@txt.sour.is on https://twtxt.net/twt/pcqworq

@xuu@txt.sour.is yikes the style sheet for blogs needs help.

@prologic@twtxt.net (#5jqioeq) Wrote up a blog post here: https://txt.sour.is/blog/xuu/2020/12/21/twtxt-auto-discovery

New Blog Post Twtxt Auto Discovery by @xuu@txt.sour.is 📝

@deadguy @prologic@twtxt.net been stewing on a discovery proto for twtxt. support for defining multiple ways to host/mirror a twtxt file. while being low tech enough to still be scriptable with basic Unix commands.

@adi @prologic@twtxt.net I’ll give it a spin first thing in the AM

Yes @prologic@twtxt.net, I will be cross-posting to twtxt.net, following the POSSE principle https://indieweb.org/POSSE . #vjdyupq

that’s it. I’m finding another twtxt client.

Atelophobia by @hecanjog@hecanjog.com [[https://hecanjog.bandcamp.com/album/atelophobia]] #links #bandcamp

@prologic@twtxt.net when its ready.. this is still in beta.

@prologic@twtxt.net It did, posting from txtnish, nice to know. #utwnv7q

@prologic@twtxt.net Testing if this will be added to the thread just adding the hashtag. #utwnv7q

@prologic@twtxt.net Txtnish adds a 28 character timestamp to the message, so in Bash: tweet=tail -1 the current twtxt.txt file and then pipe echo body=${tweet:28} to twt.

Using twt as a post_tweet_hook for txtnish to re-post to a twtxt “pod” https://github.com/jointwt/twtxt

Seems like twtxt-el does not retrieve a timeline. #emacs

Testing twtxt-el by posting this status message.

@prologic@twtxt.net yeah it would replace rice. best part is that it’s in the go build step so you don’t need to do any prep work with make.

@prologic@twtxt.net after stewing on it. I really like the idea of a wiki. throw it on the roadmap after DMs 😆

@prologic@twtxt.net (#keh22ka) maybe a custom linking method on a pod level? like can pass a template that gets translated. ex https://{domain}/wiki/{nick}/{tag} + !somepage -> https://sour.is/wiki/xuu/somepage

@prologic@twtxt.net found it!

2020-07-25T00:52:27.000000Z a new twtxt/weewiki feature: any word starting with ‘!’ will translate to an internal weewiki reference in my HTML renderer. Example: here is my !wiki_index

@prologic@twtxt.net https://txt.sour.is/twt/cwqmygq

@prologic@twtxt.net I see them is why I ask. like here #cwqmygq they use both hashtag and bangtag?

@prologic@twtxt.net 😁 that is me testing locally. does it notify you somehow when I follow?

@prologic@twtxt.net do you have any info on how the ‘!’ tags are supposed to work? are they just a different kind of hash tag?

Setting up txtnish on the Linux side of one of my computers that runs CloudReady. #twtxt #txtnish #chromeos

@prologic@twtxt.net one.. kinda sorta option would be to tailor a workflow for each of the archs.. see https://github.com/JonLundy/twtxt/runs/1568071072?check_suite_focus=true

WEBMENTION: @xuu@txt.sour.is on https://twtxt.net/twt/rxjzdwa

@xuu@txt.sour.is @prologic@twtxt.net oh.. you are.. maybe i don’t understand the issue with building?

@prologic@twtxt.net have you tried using the macos github build environment? looks like they have a windows one too.

@prologic@twtxt.net you think its time the version to get bumped? :)

WEBMENTION: @xuu@txt.sour.is on https://twtxt.net/twt/wyoav2q

@xuu@txt.sour.is @prologic@twtxt.net had to up my twt size a bit.

@prologic@twtxt.net

-----BEGIN CRYPTUTIL ENCRYPTED MESSAGE-----
l0GwFAQpx3ed+bZlcQ+pexbynFzZOm8EI/FivGbWQ16whyTkToVv8S2GSAjrsJoT
37MdaBDpoitli/f/aP130b6O6SnK/LdHHJ1DTvWgxB14sq9b4mRtk7HvYzA=
-----END CRYPTUTIL ENCRYPTED MESSAGE-----

@xuu@txt.sour.is @prologic@twtxt.net This? Fingerprint: 161c614f08e4ed4d1c8e5410f8c457e6878574dbab7c9ac25d474de67db1bdad

@prologic@twtxt.net I use https://key.sour.is/id/me@sour.is

I would need an out-of-band way to verify your public key’s fingerprint though 🤣

@prologic@twtxt.net Ok.. so using NaCL boxes. yeah its just a combo of using secretbox with a generated key/nonce. and then using the pubkey box to encrypt the key/nonce for each device.

@prologic@twtxt.net

Can we not have clients sign their own public keys before listing them on their Pod’s account?

Yeah.. we probably could. when they setup an account they create a master key that signs any subsequent keys. or chain of signatures like keybase does.

WEBMENTION: @xuu@txt.sour.is on https://twtxt.net/twt/raylsra

@prologic@twtxt.net def would be a wider discussion on preventing the pod from adding its own key to a users device list. Or using device keys to authenticate instead of user/pass.

@prologic@twtxt.net pod should probably track revocation of device keys and delete the encryptedkeys that are paired with revoked keys

@prologic@twtxt.net device gets the cypertext and uses it’s device key to decrypt one of the keys and then decrypts the cypertext.

@prologic@twtxt.net sender generates an AES key encrypts message. gets the device list for user and encrypts key for each device. sends the encryptedkeys+cypertext.

@prologic@twtxt.net for encryption. we can have browser/app generate ec25519 keypair. store the private on device and add pub to list of devices for the user on pod.

@prologic@twtxt.net 👋 I can take a stab at it when I am done with the changes I am working on.

@prologic@twtxt.net my bad.. my next one is more fun.

WEBMENTION: @xuu@txt.sour.is on https://twtxt.net/twt/xkpbn6a

@prologic@twtxt.net I see.. so using an ec25519 key as identity? and some kind of certificate to define the location of a feed? or maybe a DHT like Kademlia? TwTorrent ;)

@prologic@twtxt.net kinda like how MX records work.

@prologic@twtxt.net My thoughts on it being if they switched from a different way of hosting the file or multiple locations for redundancy..

I have an idea of using something like SRV records where they can define weighted url endpoints to reach.

@prologic@twtxt.net just an off the wall question about hashes. why not use the time+message as it was in the original twtxt.txt file? is it because it’s just not store anyplace?

also how set in stone is using user+url? vs user@domain? the latter would mean the url could change without invalidating the hash.

@prologic@twtxt.net when i get the code up to a shareable level ill ping with what i have.

@prologic@twtxt.netd so.. convert the 4 attributes in the struct to private, add getters plus some the other methods that make sense.

type Twt interface {
	Twter()        Twter
	Text()         string
	MarkdownText() string
	Created()      time.Time
    ... 
}

@prologic@twtxt.net yeah I do.

It seems a bit wonky that it imports from your packages in some places. I’m guessing that’s some legacy bits that need updates?

@prologic@twtxt.net I have some ideas to improve on twtxt. figure I can contribute some. 😁 bit more work and it will almost be a drop in replacement for ParseFile

Kinda wish types.Twt was an interface. it’s sooo close.

@lyxal@twtxt.net @prologic@twtxt.net yah. the service can have a flag for allowing non-TLS for development. but by default ignores.

are there some users that use alternative protos for twtxt? like ftp/gopher/dnsfs 🤔

My latest work over the last few days. a twtxt parser. so far looking promising. Faster and less memory than the regex version. 😁

@prologic@twtxt.net @lyxal@twtxt.net blocking http would be a good start

dither waves #shaders #GLSL #shadertoy - https://www.shadertoy.com/view/tdGBz3

Image

Image

@admin @lyxal@twtxt.net hax?

with some scripting, I could probably use my upcoming !weewiki !zettelkasten as a drop-in replacement for !twtxt, and then generate the twtxt file. however, I think I am going to keep them separate for the time being. let them both grow to serve different purposes.

@prologic@twtxt.net ❤️

WEBMENTION: @xuu@txt.sour.is on https://twtxt.net/twt/nuwjdna

@prologic@twtxt.net an added benefit of the avatar: would be the user could put their gravatar/libravatar image url like https://key.sour.is/avatar/01bc6186d015218c23dec55447e502e669ca4c61c7566dfcaa1cac256108dff0

@prologic@twtxt.net Could the config be embeded into the head comment of the twtxt.txt file and parsed out? If it also had an avatar: field that pointed to where the avatar image is located it can be almost all self contained.

New Blog Post Test Blog by @xuu@txt.sour.is 📝

FOLLOW: @xuu@txt.sour.is from @vain@www.uninformativ.de using twtrd/latest

@hecanjog@hecanjog.com has a twtxt feed yay

@lyxal@twtxt.net @prologic@twtxt.net if we edit the txt file does it update on web?

@prologic@twtxt.net the HKP is http keyserver protocol. it’s what happens when you do gpg --send-keys

makes a POST to the keyserver with your pubkey.

@prologic@twtxt.net looking through the drafts it looks like it actually used SRV records as recently as 2018 😵