Property:Contains text
D
[[Mastodon/Tech WG#TODO]]: Change the favicon i have way too many instances open in my tabs for this +
Aight i'm totally flummoxed about how to get [[Mastodon/Mathjax#Stumpted]] to load. it <really really> wants you to use the <script> tag loading method, which i would rather not do since that makes it way less portable or else requires a call-out to an external CDN, but I can't get the startup functions to trigger programmatically, and i've tried all the various things in their docs as well as some other stuff that isn't. I am tempted to just rewrite the mathstodon stuff using a react component that i was able to find but idk <@451520217139511306> u know anything else about this? +
this week is a good week to finish our rules, instance description, invite statement and make a mass coordinated announcement ♥️♥️♥️😘😘😘
in cooperation, towards information liberation!!!!!
pls join us in conversations in <#1050561894752198706> if you are so inclined 🙂
[[Mastodon/Seed Council#TODO]]
https://open.spotify.com/track/0fEasSvs63eKApKfYbi2uD +
[[Mastodon/Tech WG#Project Ideas]]: Collapse long lists of @'s in posts. see https://scholar.social/@neuralreckoning/109501954921707960 +
[[Mastodon/Mathjax]]: information in this thread about different possible implementation approaches
[[Mastodon/Hacking#Dev Environment]]: Details on setting up [[Vagrant]] for local development
[[Wikibot#TODO]]: implement n-back archiving of threads and previous posts, the parser already supports it +
[[Mastodon/Tech WG#TODO]]
[[Mastodon/Storage#TODO]] investigate [[Wasabi]] for storage 🙂 +
[[Mastodon/Tech WG#TODO]] change default email notifications to turn off new follow notifs +
[[Mastodon/Social WG#TODO]]: Once we have rules and about and more of a mod team set up we should ask to be put on fediscience's instance list [[Fediverse Indexes]] +
[[Mastodon/Tech WG#TODO]]:
- Fix semantic wikilinks in the [[WikiBot#TODO]],
- Allow `:` prefixed wikilinks to link to a page without embedding the post in that page (and then ofc stripping the colon) - this would trade off with n-back wikilinks because it wouldn't make sense to archive a series of posts at the same time as not embedding the posts into a page +
[[Mastodon/Rate Limits]]: For the record the rate limits are
- Per account and per IP: 300 requests per 5 minutes
- Upload media: 30 times in 30 minutes
- Delete posts: 30 times in 30 minutes
- Creating accounts: 5 times in 30 minutes.
https://docs.joinmastodon.org/api/rate-limits/
and a lot of methods through the UI call the API under the hood +
For the moment, until we decide on a different model, neuromatch bears the costs of the instance. Though we trust neuromatch to not abuse that power, one question we should answer as a community is whether we want to move to a fully cooperative model where the members of the instance support its costs so it is truly member-owned and operated. That would likely look like some sliding scale model where members are encouraged to pay what they can, a rough guideline based on nothing but vibes might be:
- $0/mo for students/people making <$50k annually
- $1/mo for postdocs/people making <$75k annually
- $5/mo for PIs/people making >$75k annually
though this would be something we would determine based on how many members we have and how much the instance costs to run. If we were to do this, then we would use opencollective to make our finances transparent.
We could also do a hybrid model where neuromatch kicks in some defined contribution and the members donate the rest, or if we establish a financial working group we could collectively apply for grants with neuromatch as a fiscal sponsor.
Another benefit is that by having some financial surplus we would be able to pay stipends to people who volunteer for mod duty/being on call for tech emergencies.
In the meantime though, mostly so the <#1049184335514828860> knows what it's working with: what kind of costs is neuromatch willing to bear? eg. upgrading to a 16GB ram node with 8 dedicated CPUs would be $120/month but would probably be as much scale as we need for the indefinite future. Currently we are going to upgrade to an 8GB ram node at $60/month, which is already $720 annually (just for the instance, we are also running another node that will host the wiki and loomio)
<@215274946560131074>
[[Mastodon/Finances]] +
OK [[Tech WG/Ops Diary#22-12-16 Upgrading Linode Ram]] +
[[Tech WG/Ops Diary#22-12-16 Upgrading Linode Ram]] Following the instructions to
* [[Resizing Linode]] [[Linode]]: https://www.linode.com/docs/guides/resizing-a-linode/
* [[Mastodon/Admin#Shutdown]]: https://docs.joinmastodon.org/admin/migrating/ from the migraing docs +
[[Tech WG/Ops Diary#22-12-16 Upgrading Linode Ram]] Mission accomplished. by god we've done it +
[[Mastodon/Tech WG#TODO]]: increase the size of collapsed messages: https://neuromatch.social/@PessoaBrain/109530080695260906 +
[[Mastodon/Tech WG#TODO]] [[Translation]] services! can we do something like Kolektiva does: https://kolektiva.social/@subMedia/109531114768269437 isn't there some fancy new offline/self-contained translation network thing??? I don't want to make any calls to Google from the instance but translation would be great +
[[Mastodon/Tech WG#TODO]]: why are follow buttons not present for all instances for user accounts on the notification feed? +
so! the "local" feed is the one that shows all the toots from our instance, the "federated" feed shows all toots from everyone that everyone on the instance follow. "local-only toots" are a special feature of glitch-soc where you can make your post *only* visible to the other members of the instance (in base masto you can only limit to mentioned accounts or followers). Local feed also will show you boosts from people on our instance, I'm not sure if there's a way to turn that off but if there isn't there should be bc that would be nice
[[Mastodon/Social WG#TODO]]: guides on how the different [[Mastodon/Feeds]] and [[Mastodon/Post Visibility]] work +
[[Paper Thread Bot]] the thread is here +
[[Paper Thread Bot#Resolving URLs]] https://github.com/CrossRef/event-data-reverse +
[[Mastodon/Tech WG#TODO]]: Secrets manager/password manager so we can have shared accounts better 🙂 +
OK
[[Tech WG/Ops Diary#22-12-24]] going to pull in a few upgrades:
https://github.com/NeuromatchAcademy/mastodon/pull/7
* Mathjax
* Larger Post Previews
* Default false notifications for new follows
* Autofollow accounts
And also pulling in updates from upstream glitch-soc, which in my test run means that we have to upgrade node from 14 to 16 +
ok here's [[Fetch All Replies]]: https://neuromatch.social/@jonny/109582537808504261 +
[[Mastodon/Tech WG#TODO]] Make a [[Hacking Masto]] guide including setting up a [[Mastodon/Dev Environment]] and a [[Hack Merging Policy]] +
also need general [[Ruby]] dev stuff, like a [[Masto Code Structure]] overview, which I'm feeling more and more capable of doing, and I have been doing on my personal wiki and could export over. still need to figure out how to write tests but yes I think I have gotten the big chunks of how stuff works. the [[Rails]] <-> [[Redux]] <-> [[React]] stuff still sorta eludes me, esp how it's "supposed" to be done in the program because I have more experience with react and you can do a lot of uh web app stuff with it that could also be done with rails (tho a lot of their components are a) written in the old object style which I find sorta clumsy after converting to the function style and b) actually pretty imperative for a declarative framework, which makes stuff difficult) +
the DOI thing I think will only take an hour or two more, and then yeah will be adapting some of the [[WikiBot]] code that runs on this discord already to do an agora-like thing. I've talked to flancian about this as well a bunch of times, unifying the agora bots and making a general framework for that kind of cross platform linking +
yes! this is definitely the idea with the [[WikiBot]]: capture links in situ for reminders and backlinks, but then have them be part of the wiki that can be curated down rather than just being the raw content of the post forever. Need to fix up the css for representing posts on the wiki to make them a little less obtrusive, but then yeah they can be hidden or collapsed once the page matures. +
we've talked about this as well, being able to make the bot more granularly opt-in, as well as make some simple degeneracy rules like #CamelCase == [[Camel Case]] (actually proposed that probably to no result in masto here: https://github.com/mastodon/mastodon/issues/19992#issuecomment-1363496984 ) +
<@540104559079456770> I think you should have permissions to do a code review on the PR itself, do you know how to do one of those? we should add that to the [[Hack Merging Policy]] - a guide on how to do code review.
[[Mastodon/Tech WG#TODO]] +
<@863052667857534996> do y'all have anything like this for the pyopensci stuff? like a "make a feature branch off dev, make a pull request back to dev, during a code review, look for xyz" then we will also need stuff like how to do the merge and update the instance [[Hack Merging Policy]] [[Updating Mastodon]] [[Code Review]] +
great! let me plant a bookmark on the wiki to remind me to fill in this information when I get internet at my new place & can edit it. [[Accessibility#Disabling Motion]]
[[Discord#Accessibility]] +
Ah yes, this is a [[Consensus]] thing. I just updated the wiki page to explain a bit better: https://wiki.neuromatch.io/Mastodon/Governance#Voting
Yes, you can think of it analogously to a veto. Consensus refers both to making decisions that accommodate the needs of all members as well as consent - you should not be able to be coerced into consent by a bare majority of people. In more intimate consensus processes, a single person might be able to block a proposal. That becomes untenable in larger groups for sort of obvious reasons, hence the design of this governance system to have different proportional thresholds for different types of proposals, as well as division of decisionmaking power between working groups and the general membership.
A block means something different than voting in a few ways, including that blocks should be rare: consensus sort of inverts the decisionmaking process, where care should be taken to discuss with the membership and refine a proposal before, rather than after making the proposal - so in a majoritarian system one might make a proposal and make an argument in favor, people discuss in the comments and then make up their minds there. The purpose of requiring discussion beforehand is to make more space for everyone to craft the proposal in the first place, register concerns, etc. So if people are blocking a proposal, that is a sign that the process has failed (as opposed to in a majoritarian system that the proposal has failed).
Blocking also typically enjoins the person blocking to help remedy the reason for their block: if a proposal is being made in good faith, we assume that it is meeting some unmet need by the proposer/other members. If someone blocks a proposal, that will mean that it might cause them or the organization harm, but those unmet needs still remain - so it's basically like conflict res at that point. Ofc there are exceptions like not making marginalized ppl do additional labor, etc. +
i think it would be cool if we could pay a member or two to hack on the instance!! we could take proposals on what ppl would like to see improved that guide what they are working on so we actually can resolve some of the many tiny things littering our [[Mastodon/TODO]] and maybe even one or two big things like finally doing [[DIY Algorithms]] or [[Nested Threads]] or whatnot.
[[Grants]] [[Mastodon/Grants]] +
[[Nested Threads]] Grab the parent post in this^^ +
I just had a good idea I think:
[[Mastodon/Hacks#Ideas]] [[Zenodo Social Reading]]
https://neuromatch.social/@jonny/109997725222638244
> ok you know what would actually rock: a #Zenodo plugin where you could sync a collection to an #ActivityPub outbox, and do social public highlighting that way - share a set of papers under a topic for ppl to browse, but also share your annotations and then be able to show your friends annotations over the papers you're reading if u want to
#MastoHacks #MastoDev +
@lina gathering some stuff here:
[[Mastodon/Tests]]
[[Mastodon/Tech WG]]
[[Exclusive Lists]]
[[Ruby Spec]]
[[Ruby]]
spec for home feed:
https://github.com/NeuromatchAcademy/mastodon/blob/main/spec/models/home_feed_spec.rb
for public feed: https://github.com/NeuromatchAcademy/mastodon/blob/main/spec/models/public_feed_spec.rb
that gives more examples on how feeds are supposed to be filtered
tag feed: https://github.com/NeuromatchAcademy/mastodon/blob/main/spec/models/tag_feed_spec.rb
in case we might want to add exclusive tag feeds (I personally do, self interested bc monsterdon lol)
account filter (looks like for blocks): https://github.com/NeuromatchAcademy/mastodon/blob/main/spec/models/account_statuses_filter_spec.rb
and so on
the question to me is basically how these things are structured: you have some fabricator steps that set up the conditions for the test, but then you also have these `context` statements that looks like plain English strings to me? like how do those and the `describe` fields map onto the code? like is their testing system so good that you can really just write
`it {is_expected.to be_filtered }` and that just works???? or how do you define all those.
Also u mentioned they are using a separate JavaScript testing framework? [[Jest]] ? that looks like it's configured here: https://github.com/NeuromatchAcademy/mastodon/blob/main/jest.config.js
and doesn't cover the glitch flavours, just the mastodon folder. it seems like that is for testing the appearance and function of the JS? not sure how it interacts with the ruby tests.
So for the sake of us learning to write tests here, I don't think we need to aim for full adversarial coverage of the feature, but just get the basics: "when an account is in an exclusive list, their posts dont show up on the home feed." judging from the other tests, it looks like we'll have to also test for boosts from them, but since it's not like a block we don't have to test a bunch of edge cases like interactions.
re: [[PaperBot]] we also had talked about [[DOI Search]] and resolving urls to DOIs and embedding them in the [[JSON-LD]]. the DOI Search should actually be super easy if we do it by altering the hashtag regex. doing it by making it an additional search type would take a little more, as would doing automatic dereferencing, but all those sound like fun to me +
[[Mastodon/Social WG]]
[[Moderated Instances]]
new reports include a few spam accounts from mastodon.top, looks like a big general instance, though I can't actually see it because their certs are broken. going to ban the reported accounts and put the instance on notice, raising here in case anyone has any input on the instance +
ok [[Tech WG/Ops Diary#23-03-29]]: going to pull in dev branch and deploy to tha server: https://github.com/NeuromatchAcademy/mastodon/pull/13
I thought this had fetch all replies in it but i need to rebase it and also now that we know how to write them i want to write some tests....
It does have [[Better Code Blocks]] in it from <@485848481995423755> : https://github.com/NeuromatchAcademy/mastodon/pull/9
wish e luck +
QUESTION isn't [[NMA]] coming up? and maybe [[NMC]] at some point???? how can the instance support that????? can we make some cool pages for people to share their code or whatever, index the talks in a conference, etc???? I have posted a few ideas before in the neuromatch slack but wanted to pose this generally to the instance and see if we can yno use that as a motivating reason to do some masto hacking
tagging the neuromatch ppl I know are on the discord <@540104559079456770> <@1050459047582781480> <@215274946560131074> <@651480296415297553> and who else sorry I don't have a list
[[Mastodon and Neuromatch]]
[[Neuromatch and Mastodon]] (idk lol I'll just redirect one page to another)
[[Infrastructure]] +
So [[Masto Docs]] [[Rate Limiting]] is controlled by [[Rack Attack]]
- https://github.com/NeuromatchAcademy/mastodon/blob/main/config/initializers/rack_attack.rb
- logging: https://github.com/NeuromatchAcademy/mastodon/blob/main/config/initializers/rack_attack_logging.rb
- docs: https://github.com/rack/rack-attack#throttling
The throttling rules differentiate between authenticated and unauthenticated API requests, so I would want to raise the authenticated ones. Unfortunately the logging does not actually tell me what *kind* of throttle is being hit (eg. the authenticated api vs token api vs paging api, etc.) and the throttling rules definitely overlap with one another... so I'm going to make that change first - log the kind of throttle that is hit, and then come back later and actually adjust values +
sidenote I just looked at the channel member list and holy crap wikibot are you still alive [[WikiBot Checkup]] +
[[Tech WG/Ops Diary#23-05-19]]: We are attempting to [[Upgrade Mastodon]] to 4.1.2 and also merge in [[Exclusive Lists]], wish us luck.
We begin with what is probably the worst way to do it based on the conversation in <#1049184335514828860> , specifically:
* main -> dev,
* upstream -> dev,
* dev -> pr,
* pr -> dev,
* dev -> main
https://discord.com/channels/1049136631065628772/1049184335514828860/1109355378178789396 +
[[Tech WG/Ops Diary#23-05-19]]: [[TODO]] check out how user-specific settings were reformatted re: config/settings.yml. See also https://discord.com/channels/1049136631065628772/1049184335514828860/1109359788602707988 and https://discord.com/channels/1049136631065628772/1049184335514828860/1109360742433558548 +
ok [[Tech_WG/Ops_Diary#23-05-19]]: We conclude this evening having gotten all the way to the last stage where we pull back to main and then deploy. see preceding chat logs for history and also for the millionth time [[WikiBot#TODO]] implement archiving ranges of posts +
[[Tech WG/Ops Diary#23-05-22]]: What up it's me and we're about to deploy [[Exclusive Lists]] and update to 4.1.2. starting by making a linode backup and then following instructions on https://github.com/mastodon/mastodon/releases/tag/v4.1.2 +
yes agreed we should make a wiki guide to updating the instance.
I need to log off for the day bc am visiting friends, but in short it's
- start backup in linode (reminder to also set you up with an account there and add that to [[Tech WG#TODO]] [[Tech WG/Onboarding]] docs
- do all the git work that we're familiar with to prepare the new deploy
- check upgrade notes, usually on base masto release notes, to see if anything out of the ordinary is required
- pull down new changes to server
- potentially recompile yarn assets
- potentially run database migrations
- restart mastodon-* services
and that's usually it. downtime is usually on the order of a few seconds (and we could explore how to make that zero, I think that's how the pros do it with like multiple servers) +
- solidarity with the fedi: to survive this and future attempts at corporate capture, it is important to act together. many instances have already committed to defederation, and even if it is symbolic, symbols are important.
- being defederated by other instances: we are currently in the good graces of some very protective instances through prior social ties and our promise to be a good actor in the space, these include instances with many leftists, queer people, and others that would likely defederate from us if we were to federate with meta. this would put NMS in a more homogeneous, corporate nice space in the fedi.
- communication with threads users: some people probably want to communicate with some threads users, and this might cause dual accounts, as you say. defederation would potentially split ties. I will hold my thoughts on this for now since I'm just trying to list the relevant factors.
what else? listing the concerns, we could then see how each impacts a decision to limit/defederate based on what each decision does.
I am traveling and AFK until Monday, but I think it would also be good to make a [[Defederation from Meta]] wiki page to keep more organized notes. forgot to tag the previous post. +
Also saw you boosted this, seems like a great idea for [[Translation]]
https://frontrange.co/@raineer/110959799057137486 +
[[Tech WG/Ops Diary#23-09-19]] - Upgrading to 4.2-RC2 while trying to write the [[Upgrade Mastodon]] docs. currently doing the merge from upstream and noticing db migrations for exclusive lists failing, but not fixing now because that would put the fixes in the `glitch-soc-main` branch which is a local mirror of upstream which gets force pushed. Seems like that should happen in the `merge-upstream` branch after upstream is merged +