Maiyannah BishopI continue to be amused by people who want to discuss changes to a social network on anything but that social network.
All of the people who would be affected by any proposed change to the social networking standards are _right here_ talking on this thing _right now_. Any alternative is just going to be a subset of those people at best. If you want to discuss changes I'm going to consider at all, discuss them here, or don't bother.
sungo@maiyannah But they might get involved in the conversation before there's a common front by the admins/devs! They might have an opinion before the admins/devs are ready to hand down an edict! *eye rolling so hard they stick*
@maiyannah I remember when a decision to create an admin-only instance and to use Discourse heavily happened a couple months ago. It was entirely about controlling the narrative. There was a distinct desire for a lack of transparency about the ongoing operations of instances.
Well, and making sure the messages didn't propagate to the GS side of the fediverse because god help us if the OLD TIMERS had thoughts.
Well, and making sure the messages didn't propagate to the GS side of the fediverse because god help us if the OLD TIMERS had thoughts.
@sungo By splitting the community like that they empower a few at the cost of the many. It's literally the oppression dynamic. I reject it.
Nolan Lawson@maiyannah @sungo Just for context, I'm the one who spearheaded the admin-only instance (which is basically dead now BTW). My original goals were:
- move meta-talk off of the timeline; it was consuming all discussion
- try to get admins to at least talk to each other
- originally it was admin-only, but I relaxed this and even added a non-admin as an admin (insert Bertrand Russell reference) to avoid any appearance of elitist cabal
@nolan @maiyannah while I get that meta-talk was insane back then, consuming all discussion was actually important because folks were defining the culture of the system. Folks could always mute you all as I did from time to time. When the admins disappeared, the users lost their voice and their say in the culture you all were creating.
@sungo @maiyannah Discourse forum is open; do you see this as a better alternative w.r.t inclusiveness? I agree my idea of "admins only" instance was too secret cabal-y.
@maiyannah @sungo I dunno, I kinda feel like it's too ephemeral here, and lack of searchability makes it hard to keep a paper trail (dunno if postActiv's UI is more amenable to this). I doubt Twitter's devs use Twitter to discuss dev stuff, so to me it seems fine to use some other communication software for dev/community talk. Already a lot of it is on GitHub.
@sungo @maiyannah My point was that if you're building a communication platform you don't have to use that platform for every kind of communication. E.g. I'm pretty sure Mozilla Thunderbird devs use IRC to chat and don't feel like they need to use email for everything. But your points elsewhere about dogfooding are well-taken. :)
@nolan @maiyannah But Thunderbird isn't an application for real-time chat. If you're a thunderbird dev and you're not using thunderbird to send email, you're doing it wrong. If you're an IRC dev and you're not using IRC, you're doing it wrong. you are asking your users to find all your problems for you and, then, because you're not using it that way, you end up with responses like "that's not an issue I have. Ticket closed" which has happened. FFS, the "make columns wider" conversation went exactly like that. "I like the column size. Issue closed".
@sungo @maiyannah Kinda feels like we're talking about a few different things here, but I think the project devs use Mastodon plenty and that's not so much the issue. I happen to agree with Eugen about the column size thing but for the general issue you raise I think the situation could maybe be improved by a foundation/governance model which is a topic I've broached: https://discourse.joinmastodon.org/t/mastodon-project-governance/215
@nolan @maiyannah You know how you handle something like column sizing? You make it configurable and make the default something the devs think is ok. Then, if I as the user or admin want something terrible, I can have it but by default the system is how you think it should be. It's all about choice and user/admin freedom.
@sungo @maiyannah Sure but every additional feature makes it harder to maintain the project. I've run into this in literally every OSS project I've been involved in that reached a certain level of popularity. Every user has their pet/idiosyncratic feature request, and if you say yes to every one, then the project quickly looks like Homer's car with 100 bells and whistles and configuration options. Also folks are always free to fork and modify.
@maiyannah @sungo If I understand what you two are saying, it sounds like there are two issues here: 1) folks feel uncomfortable with Eugen making unilateral decisions, and 2) Mastodon has a hugely outsized influence on the fediverse because it's so popular. #1 feels to me like it's solvable with governance, #2 seems solvable with standards and compat effort (similar to browsers, e.g. when one browser has huge market share and thus influence on web ecosystem).
@maiyannah @sungo Hm yeah this is a problem I know well from the browser space (try convincing Chrome they need to change their impl when they have 60% market share, or IE6 back when it had 90% 😉).
It seems to me that having a common test suite and making it as part of the CI testing for all OStatus projects would be a big step forward. Browsers are doing this now but should have done it decades ago.
@maiyannah @sungo If you want to collaborate on a PR to Masto to run the test in CI, let's talk. :) I have commit access to Masto, although of course I'll run it past Eugen first. But I think compat is important because it's a big part of the whole "anybody can make an instance, it's an open protocol, it's all one big open-source fediverse" story.
gled => Moved@nolan @maiyannah @sungo
adding @wonderfall and @Hexalyse as we were talking about that exactly at the moment, mentionning the 'no favorite' bug in 1.4.6.
The point is that CI and unit test importance does not seems to be too much of a deal from the lead dev team, probably because of the lack of experience on that matter and why it's so important ?
@Hexalyse @wonderfall @sungo @maiyannah @nolan Nolan, I can also probably help on the CI part, but I won't loose time to get a 'rejected because I'm hungry' from Eugen ;)
@gled Cool, maybe we should open a GitHub issue to discuss? Are you already on GitHub or in Discord or somewhere? :)
@nolan I'm gled-rs on GH, no discord for me ;)
Feel free to mention and I'll bring my flipflops to you ^^, I'll probably draft a first test on my fork this weekend, to at least secure mastodon.host from regressions ;)
@gled I'm nolanlawson on GitHub, I might open an issue but need to know where to start, just poking around for now. :)
on the dev side, we need unit tests ( which is gonna be a huge work ) for everything.
On the CI part, we can probably leverage Travis as it's free for OS projects.
@nolan but that also mean no merging of a PR if the tests don't pass. No exceptions, whoever is the committer.
And no silencing of tests also to make them pass ;)
beatrix bitrot@bea @nolan Thanks for the offer, but I'll probably do it here: https://github.com/gled-rs/mastodon as it's already auto-deployed on my server so I just need to add the 'check CI before deploy' and I'm always on a stable, fully functional release, which is fun to develop ^^
i am over here
https://github.com/glitch-soc/mastodon
with a few other people having fun breaking the software and adding small stuff so far
how are you doing your automated deploys?
i set that up in a crude way for myself but i would like to incorporate CI as well and i think it would be troublesome with the current setup
@bea @nolan starred, gonna keep a look on your fork ( which makes me think that if I'm a bit bored this weekend, I may introduce a new feature for you: random toots character limits ) :)
Regarding deployment, I have a global ansible based orchestration server that manage my infrastructure, and mastodon.host by extension.
@bea @nolan I tend to recall that Travis has callbacks no ?
Not sure about that though, it's been a while ( there is on Jenkins, used every day here, it's part of the orchestration ).
I don't admin anything anymore on my servers by hand, it took a bit of time to setup, but now it's so easy, even for testing that new shiny thing that just got released, I just code a quick playbook, integrate it and wait for the server to boot perfectly configured ^^