MOVED to @>mastodon unthreader
gargron please increase the character limit
gargron please increase the character limit
Toby Inkster 
@chjara It's fairly easy to patch Mastodon to change the limit, but I agree it would be nice to make it configurable.
@tobyink it's intentionally hard to do and hardcoded in as many places as possible to discourage doing it lol
Hera [Yaseenist] Harper
Verilog Connoisseur 
@[email protected] @[email protected] @tobyink it's also a lot more demanding and scales much worse than basically any other software
I think the only reason people are still using it is cuz it was first
I think the only reason people are still using it is cuz it was first
LisPi@s1ide @chjara @tobyink @alfredohno Pretty much. #Rails isn't exactly famous for its amazing performance and even the best #Ruby runtimes like #TruffleRuby propose around 5x performance at best which... just isn't good-enough particularly when Rails itself /doesn't/ get 5x improvement from that.
Scaling wasn't a primary consideration and tacking it back on post-facto is a lot more complicated.
Jared White (ResistanceNet ✊)@lispi314 @s1ide @chjara @tobyink @alfredohno What is missing from this analysis is that Rails makes it *really* easy to *build* really fast…essentially letting a solo dev build out an entire platform. I'm not in the least bit surprised Mastodon was built with Ruby, and I'm not at all confident we'll see an obvious non-Ruby alternative emerge any time soon.
@jaredwhite @lispi314 @s1ide @tobyink @alfredohno uhh i'm posting this from akkoma which is written in elixir
@chjara @lispi314 @s1ide @tobyink @alfredohno I never said there are *no* alternatives to Mastodon for ActivityPub software.
@jaredwhite @[email protected] @lispi314 @tobyink @[email protected] you said there were no obvious non-ruby alternatives
Pleroma and its forks are written in elixir, misskey and its forks are written in nodejs. They both have more features than mastodon and are significantly lighter weight. I'm currently running my instance (calckey, a misskey fork) on a ~10 year old mid range laptop, and it works great. That's something I would not be able to do with mastodon.
Pleroma and its forks are written in elixir, misskey and its forks are written in nodejs. They both have more features than mastodon and are significantly lighter weight. I'm currently running my instance (calckey, a misskey fork) on a ~10 year old mid range laptop, and it works great. That's something I would not be able to do with mastodon.
@s1ide @chjara @lispi314 @tobyink @alfredohno again, *obvious* alternatives as in the mainstream option can now be jettisoned in favor of some net-gain altenate option. I don't see any indication that "Calckey" is there. Maybe in a few years! Let's check back then! 😅
@jaredwhite @[email protected] @lispi314 @tobyink @[email protected] but there is a net gain. Yes, it's less mature software but it has more features (emoji reactions, user defined themes, post translation, admin-configurable post length) and runs/scales better. I think the only reason why misskey based instances aren't more popular is because migration from mastodon isn't there yet and not everyone knows about it. Also I don't see why you put "calckey" in quotes unless you're just trying to be condescending.
@s1ide I'd imagine migration difficulty would be responsible for a large part of it.
@lispi314 most likely, I had enough trouble migrating from misskey to calckey
The other issue is that misskey currently doesn't support the masto API so you can't use apps like megalodon, tusky etc. It is a PWA so it can be "installed" on most devices and still be fully featured as a mobile app, but it does have some disadvantages
I've been following calckey development and both these issues are being worked on as we speak
The other issue is that misskey currently doesn't support the masto API so you can't use apps like megalodon, tusky etc. It is a PWA so it can be "installed" on most devices and still be fully featured as a mobile app, but it does have some disadvantages
I've been following calckey development and both these issues are being worked on as we speak
@chjara @s1ide @jaredwhite @alfredohno @tobyink I wonder if #Elixir's superficial resemblance to #Ruby played a role in that.
Superficial because good Elixir (basically, #Erlang & #BeamVM) performance will very much not result from reusing the same kinds of patterns.
Nat Bennett@lispi314 @chjara @s1ide @jaredwhite @alfredohno @tobyink Elixir's resemblance to Ruby is not accidental & it's pretty deep IMO. The languages share a lot of values, including developer productivity.
I'd, uh, kinda be surprised if it was *Ruby* causing any Mastodon scaling issues though.
@nat @chjara @s1ide @jaredwhite @alfredohno @tobyink A lot of it is probably on Rails specifically, but at the same time Ruby is largely still interpreted with a GIL to boot, is it not?
@lispi314 @chjara @s1ide @jaredwhite @alfredohno @tobyink Sure but I'm skeptical of the idea that on-host processing time is a significant factor in application performance at the scale of a Fedi server. Generally it's architecture & esp. data access that gets you.
Every story I've ever heard about how "Rails is slow" that I got more details on later turned out to involve just horrific things with the database or like, a failure to tune something basic.
@nat @chjara @s1ide @jaredwhite @alfredohno @tobyink I couldn't say whether that's not the case here or generally, I have no personal experience with the #Ruby ecosystem and only learned what little I know about it when looking into claims of #Mastodon badly under-performing.
I wouldn't be inclined to dismiss #ORM-related ugliness being responsible for a large part of it.