# I am the Watcher. I am your guide through this vast new twtiverse.
#
# Usage:
# https://watcher.sour.is/api/plain/users View list of users and latest twt date.
# https://watcher.sour.is/api/plain/twt View all twts.
# https://watcher.sour.is/api/plain/mentions?uri=:uri View all mentions for uri.
# https://watcher.sour.is/api/plain/conv/:hash View all twts for a conversation subject.
#
# Options:
# uri Filter to show a specific users twts.
# offset Start index for quey.
# limit Count of items to return (going back in time).
#
# twt range = 1 148
# self = https://watcher.sour.is?uri=https://xandkar.net/twtxt.txt&offset=148
# prev = https://watcher.sour.is?uri=https://xandkar.net/twtxt.txt&offset=48
@lyse ah, right - forgot to explain that one - that is indeed the unix epoch timestamp, of the latest message that nick was seen in.
The better thing is probably to pull the desired nick from metadata comments, but not everyone has it and usage frequency is still interesting on its own.
Tracking current/correct nicks was getting confusing, so I added a nick usage frequency count pass to the crawler, which also corrects a URL's nick to the one most-frequently mentioned for that URL. The collected nick data per (encoded) URL is here: https://xandkar.net/twtxt/nicks/seen/
@movq logrotate feels like the right solution. Just need the client support and a convention for where to host the archives.
Added deduping by URI, which trimmed the unverified set to 1244 and verified to 583: https://xandkar.net/twtxt/ . Still more refining to do, but it's starting to look more reasonable now. H/T: @lyse
Perhaps because I'm not limiting to just twtxt.net? I seeded with initial peers manually and from there keep spidering out to mentions. Also, there may still be dupes which differ only in nick and/or some alias URI that leads to the same place - need to design a de-dupping pass...
If a peer's URI is downloaded with 200 and at least one message is parsed from it - a peer gets the "verified" badge.
Hey @prologic! Been busy with work and family and what not. How's it going? Also see my previous message.
Sorry if I missed replying to you. Besides time constraints, my client is currentl stupid and does not accumulate messages - if you trimmed your timeline - I lost the trimmed messages forever. I _might_ work on improving this over the winter vacation.
Finally added validation to discovered peers! The list of now-VERIFIED, discovered peers, with at least one valid message is now at 703 (out of the 1449 unverified - a lot of garbage mentions are out there). https://xandkar.net/twtxt/peers.txt
@prologic I know - my list is completely unverified - just parsed mentions and follows. There're completely bogus entries, like twitter and google. I know what needs to be done, just haven't prioritized doing it yet.
@hecanjog RE: "way out of capitalism" - it's really interesting to read the slightly-expanded form of what you mean. People tend to have their own implicit associations with terms (like "capitalism", "anti-capitalism", etc.) and end-up arguing right past each other. There seems to be a lot more overlap than disagreement over the problems with the structures supporting modern life. Although that might not be too surprising. What is somewhat more surprising is that the final ideal seems to also be the same - honest relationships. The universal disdain seems to be toward systemic scams, which capitalize on accidental patterns and betray essential goals.
Xers and Mils are fast on their way to becoming the proverbial Boomers - the parrots of cliches. Could it be just people in general and nothing particular to any one generation?
@prologic This data is pretty low quality right now. I'll add verification of successful download and parse before submitting it. (#2yfxkra)
Which now makes me wonder - what other deep and interesting ideas got a bad rep by vice of popular meme-ification?
I used to think of meditation and "mindfullness" as newage hippie bullshit, until I accidentally came upon it's description in Siegel's book - turns out it's actually great.
@prologic Nice! Good call on User-Agent too. (#fu3pmea)
Crawled ~1100 peers so far. Naively parsed from mentions and unverified, so many are invalid, inactive and/or near-duplicates. I'll update this list periodically: https://xandkar.net/twtxt/peers.txt
@prologic Testing my new crawler... Oops - should probably run it as a different user than myself ... :) :thought_bubble:
@anth +1 on plain text readability! twtxt is a completely different beast without it.
@jlj RE: tags - doesn't have to be a single word per se, just that it is made of words from a human language, rather than some hash digest + platform-specific search URI
I'm starting to think that it is not just still the case today, but has always been that way and cannot be any other way. It is just that what we classify as "religion" is but narrow sample of the broader phenomenon.
In earliest states there was no such thing as religion. What we call religion today, was just reality, just normal life things - state and religion were indistinguishable.
@niplav Thanks for the slatestarcodex link - interesting!
@jlj Please just use a human-readable #tag. Otherwise it is a terrible UX for people reading in plain text.
@jlj Hear, hear! Over the years, Thunderbird has consistently been the client I hate the least.
Philosophy always ends predictably.
@prologic Are "distributor tags" something different from above? Where can I read about them?
Simple, low-dependency viewing and parsing is what makes both twtxt and IRC so fun.
@prologic Yeah, I'm not crazy about the conversation grouping stuff. Simple #tags are good-enough and don't clutter the txt-only view of the timeline.
@prologic Probably something respecting individual rate-limits would need to be done... :thought_bubble:
@prologic Haven't thought of anything particular interesting yet, which is why I'm asking :)
@prologic Ah! Last-Modified is a good call. What are you thinking WRT to making it smarter?
For a more IRC-like experience :)
General twtxt extension idea: instead-of|in-addition-to twtxt.txt file - we could have a directory for topics/channels with a file per topic: http://foo/twtxt/${topic}.txt
I want to add polling to tt and so far just thinking randomized intervals + ETag check.
@prologic Somewhat lazy question: what is your polling strategy for such large numbers of twtxt.txt files? Interval + ETag check? Something more|less clever?
RSM's antiglossary is still up and is an interesting read: https://stallman.org/antiglossary.html - Can't say that I'm onboard with everything, but it's clear that he is actually thinking, which _is_ interesting, as opposed to the people shouting him down.
@jlj Seems more like some personal vendetta than rational engagement.
[2/3] added a custom upload alias, added sort reversal, simplified install - now building a stand-alone executable:
\n added a custom upload alias, added sort reversal, simplified install - now building a stand-alone executable:
[1/3] Made some tt improvements this weekend: fixed timezone handling, added subcommands, separated reading and downloading commands/stages,
\n Made some tt improvements this weekend: fixed timezone handling, added subcommands, separated reading and downloading commands/stages,
mpd/mpc hack to loop a segment within a long audio file - while it's playing, execute from shell: while :; do mpc seek $BEGIN_TIME; sleep $DURATION; done
no doubts <=> no thinking
@prologic Joined. Oh, duh! I see the "Ordering" section now... Thanks! :)
@prologic How is the AoC leaderboard ordered, by completion time? I see 3 of us have all the coins, but different points.
A lot of random new things to get used to... evil mode is helping. I have to say though - Elisp is far more motivating than VimL!
Hmmm - having a hard time generating a minimal example...
Whoa - just noticed that git detects that a file was copied and marks the operation as so.
Poetry: "As the number of lines of code in your software shrinks, the more skilled you have become and the less your software sucks." - had a re-read of this today after referring a colleague to the site: https://suckless.org/philosophy/
The main idea is to not have a master "following" file, but to choose a desired subset on each run, for example: tt <(grep foo following)
@deadguy Have basic instructions and CI now: https://github.com/xandkar/tt . Note that it is only a reader at the moment. Takes a single argument: a file with lines. Your info it tries to read from the first line of ~/twtxt-me.txt - this is not meant to stay - will change to a more-thoughtful config system later :)
@prologic Moreover, using a URI as the host/reference-point of a _conversation_ (which differs from a mention) - feels like it detracts from the goal of decentralization. I understand the desire to disambiguate context, but it seems to be an overkill with more drawbacks than benefits.
@prologic Checking-out how you guys used hashtags to track conversations - it is undoubtedly clever! It doesn't quite feel right to me though - too-heavyweight, too-specific and clutters the text-only reading experience. Plain old hashtags could still accommodate this use-case without these drawbacks (just postfix a hashtag with a unique-enough ID #for-example-hg7c3n).
There's something noticeably distinguished about the personalities I see involved in the circles of Lisps. A certain intellectual bravery and broadness of curiosity that just feels different from anywhere else, including other FP circles (some examples: Erik Naggum, Francois-Rene Rideau, range of topics and personalities at http://lisp.nyc/).
@deadguy Sweet! It's very much half-baked at the moment though. No setup or usage instructions, lack of features (other than downloading and displaying a timeline), etc. I'll post a heads-up when it's a bit more usable. There're a bunch of TODOs at the top of the file.
peeve: use of double quotes in shell scripts where no expansion is desired
A nice backlog for rainy weekends :)
Have a bunch of improvement ideas too. Like virtual "channels" based on hashtags and virtual timelines based feed-set combinations.
It is very much WIP, but already useful. Decided to show as an IRC-like log with RFC2822 timestamps.
@lucidiot He really had a way with words! In one of his C++-attacking posts he mentions having made a living as a writer at one point, which isn't at all surprising.
@kas: (re: f) neat idea! I often wish for something like that, but then convince myself that "ok... typing {print $n} is not _that_ bad" (even if it kind of is... lol)
2) what is truly important for the long haul is less-so hairball artifacts of implementations, and more-so clear communications of ideas through small examples and abstractions.
1) it is useful to be conscious of our non-freedom;
From there, (at least) 2 more things follow:
It follows that 100% freedom is impractical, but must be nonetheless be aimed at.
Most who say "I'm democratizing X", really mean "I'm honeypotting surfs". Implementing is freedom, using is slavery.
Similarly with project planning/management. I prefer a single text file as the master source of organization, while a ticketing system is a communication tool.
As a personal tool, I still find a dead-tree planner most useful. An electronic calendar seems best as a group communication medium rather than a personal organization tool.
@lucidiot systems to get rid of ideas - I'm stealing that one :)
The Freehackers Guild. Featuring esoteric Unix rites and master belts of parenthesis.
Whenever I grasp an existing idea for the first time, I tend to feel a sort-of panicked embarrassment - "OMG! Everyone knew this, how could I have been so dense?!"
Program evaluation is topological sorting. An in retrospect obvious thought.
@johanbove I registered twtxt.club - plan to put a simple twtxt crawler and aggregator there when I get a chance.
Another emerging heuristic: people that tend to lean heavily on "X is more empirical than Y" as an argument, lack awareness and/or honesty.
I often feel that way about Taleb, Wolfram and Musk. Some things I see that they're spinning, but only because I know the things, but I can't tell about other things, while a car salesman is always obvious, even if I know nothing about cars.