Gregory's Server:fse::hacker_f::hacker_s::hacker_e::fse:
:smb64_a::smb64_l::smb64_s::smb64_o:
:revolvertan::mgsgb_r::mgsgb_e::mgsgb_v::mgsgb_o::mgsgb_l::mgsgb_v::mgsgb_e::mgsgb_r::revolvertan:
:smb64_a::smb64_l::smb64_s::smb64_o:
:pleromatan::lazer_p::lazer_l::lazer_e::lazer_r::lazer_o::lazer_m::lazer_a::pleromatan2:
Okay, we're back! Box locked up. I would like to get Revolver out the door so that this machine can be retired.
Hilariously, we were only half-up for a while because I forgot to bring up Revolver and nginx sends some of the incoming reqs to Revolver and some of them to Pleroma, like we do with webvac, but webvac starts automatically and Revolver I didn't write an init script for because it's being tested and I'm just running that shit from a screen session (like a :chad:).
Now that I have tamed the part of ActivityPub that was causing me some trouble, I have one or two things to test, and then from there the last bit is hopefully just a matter of doing a bunch of dumb shit to support the frontend. That's a non-trivial amount of work but it's also not difficult work, but I've been doing mostly backend/engine-y stuff because computing is fun and webshit is boring. Anyway, fingers crossed, we may be in the home stretch and I could not be more excited, though hopefully the infrastructure doesn't come crashing down around me before it's able to .
A fun bit, there's a small bug in https://freespeechextremist.com/rvl/stats : the first req after it comes up kicks off a scan of rstore[1], which is all of the data in the system, which apparently takes forever to read, because there's no index yet. (It's about 1GB split into files less than 8kB.)
On another note, somewhat related, during Soapbox's reply brouhaha, I had remarked that FSE remains fully behind the Pleroma project, because the Pleroma developers are good people and (most importantly for the network) they care about the network and ensuring that the fediverse interoperates smoothly. This has not changed, and FSE will continue backing whatever fork is good for the network, which continues to be mainline Pleroma. When Revolver is out, there are some changes that will help, especially smaller instances: I've gone to some effort to make sure the object proxy functionality is usable, meaning that the entire network should have an easier time federating, and I have some planned tweaks to Pleroma to support alternative intake methods for AP objects as well as the media proxy, which should alleviate load on the network as a whole and even allow temporarily dead instances to survive. Any patches I write will be sent upstream to Pleroma, because that is the project where the developers have, repeatedly over the course of years, demonstrated that they care about the network and will guide the project along the course that is best for the fediverse.
More software is good for the network: Honk and Pleroma and PeerTube and Pixiv and even Mastodon, so the forks are fine. Unfortunately, the developers of two of the recent forks of Pleroma appear to have shat the bed completely on the simple principle that all of these instances are useless in isolation and what matters is the network. Not only that, but they have built their software on minimal tweaks to Pleroma's code while trashing the people that did all of the work. Mainline Pleroma has prioritized operating properly with existing deployments, keeping code in working order, even bending over backwards to support Mastodon's loopy nonsense, to stay compatible. This is one of the design principles of the project, per lain: "That’s something I’d recommend to other fediverse projects, too: Write compatible APIs." ( https://medium.com/we-distribute/blushy-crushy-fediverse-idol-a-chat-with-lain-about-pleroma-4ff578b99752 ). Mastodon is terrible and repeatedly done things that break everything that isn't Mastodon while simultaneously pretending it is the entire network, and I have not ever seen vitriol from the Pleroma devs about Mastodon, while the people using Pleroma as the basis for their projects have spent more time trying to spew bile on Pleroma than they have spent hacking. Everyone could do with a little cooling down and being respectful of the people that built this world we inhabit: without all the work that went into Pleroma, most of us wouldn't be here. It is certain that Akkoma and Soapbox would not exist, and to exploit Pleroma this way is disheartening to see from people that are ostensibly hackers. I have more to say, but I had better not, so lemme try to keep it positive:
:revolvertan::oppaiheart::pleromatan::mgsgb_4::lazer_e::lazer_v::lazer_e::lazer_r::hoshii_exclam:
:smb64_a::smb64_l::smb64_s::smb64_o:
:revolvertan::mgsgb_r::mgsgb_e::mgsgb_v::mgsgb_o::mgsgb_l::mgsgb_v::mgsgb_e::mgsgb_r::revolvertan:
:smb64_a::smb64_l::smb64_s::smb64_o:
:pleromatan::lazer_p::lazer_l::lazer_e::lazer_r::lazer_o::lazer_m::lazer_a::pleromatan2:
Okay, we're back! Box locked up. I would like to get Revolver out the door so that this machine can be retired.
Hilariously, we were only half-up for a while because I forgot to bring up Revolver and nginx sends some of the incoming reqs to Revolver and some of them to Pleroma, like we do with webvac, but webvac starts automatically and Revolver I didn't write an init script for because it's being tested and I'm just running that shit from a screen session (like a :chad:).
Now that I have tamed the part of ActivityPub that was causing me some trouble, I have one or two things to test, and then from there the last bit is hopefully just a matter of doing a bunch of dumb shit to support the frontend. That's a non-trivial amount of work but it's also not difficult work, but I've been doing mostly backend/engine-y stuff because computing is fun and webshit is boring. Anyway, fingers crossed, we may be in the home stretch and I could not be more excited, though hopefully the infrastructure doesn't come crashing down around me before it's able to .
A fun bit, there's a small bug in https://freespeechextremist.com/rvl/stats : the first req after it comes up kicks off a scan of rstore[1], which is all of the data in the system, which apparently takes forever to read, because there's no index yet. (It's about 1GB split into files less than 8kB.)
On another note, somewhat related, during Soapbox's reply brouhaha, I had remarked that FSE remains fully behind the Pleroma project, because the Pleroma developers are good people and (most importantly for the network) they care about the network and ensuring that the fediverse interoperates smoothly. This has not changed, and FSE will continue backing whatever fork is good for the network, which continues to be mainline Pleroma. When Revolver is out, there are some changes that will help, especially smaller instances: I've gone to some effort to make sure the object proxy functionality is usable, meaning that the entire network should have an easier time federating, and I have some planned tweaks to Pleroma to support alternative intake methods for AP objects as well as the media proxy, which should alleviate load on the network as a whole and even allow temporarily dead instances to survive. Any patches I write will be sent upstream to Pleroma, because that is the project where the developers have, repeatedly over the course of years, demonstrated that they care about the network and will guide the project along the course that is best for the fediverse.
More software is good for the network: Honk and Pleroma and PeerTube and Pixiv and even Mastodon, so the forks are fine. Unfortunately, the developers of two of the recent forks of Pleroma appear to have shat the bed completely on the simple principle that all of these instances are useless in isolation and what matters is the network. Not only that, but they have built their software on minimal tweaks to Pleroma's code while trashing the people that did all of the work. Mainline Pleroma has prioritized operating properly with existing deployments, keeping code in working order, even bending over backwards to support Mastodon's loopy nonsense, to stay compatible. This is one of the design principles of the project, per lain: "That’s something I’d recommend to other fediverse projects, too: Write compatible APIs." ( https://medium.com/we-distribute/blushy-crushy-fediverse-idol-a-chat-with-lain-about-pleroma-4ff578b99752 ). Mastodon is terrible and repeatedly done things that break everything that isn't Mastodon while simultaneously pretending it is the entire network, and I have not ever seen vitriol from the Pleroma devs about Mastodon, while the people using Pleroma as the basis for their projects have spent more time trying to spew bile on Pleroma than they have spent hacking. Everyone could do with a little cooling down and being respectful of the people that built this world we inhabit: without all the work that went into Pleroma, most of us wouldn't be here. It is certain that Akkoma and Soapbox would not exist, and to exploit Pleroma this way is disheartening to see from people that are ostensibly hackers. I have more to say, but I had better not, so lemme try to keep it positive:
:revolvertan::oppaiheart::pleromatan::mgsgb_4::lazer_e::lazer_v::lazer_e::lazer_r::hoshii_exclam: