<iconoclasthero>
hi, quick questions: is wayback going to be able to support X over SSH (ssh -x / ssh -Y)? is wayback going to be compatible with e.g., GNOME 50 on an Ubuntu-like distro?
<iconoclasthero>
apparently google thinks it should send me news articles about these things (not not the SSH stuff—yet) and while I understand the above, I don't know what " It is essentially a stub compositor which provides just enough Wayland capabilities to host a rootful Xwayland server." means.
<ToyKeeper>
iconoclasthero: If you're using Wayback locally and you open a terminal and ssh -X to another system, you should be able to run X11 programs over that ssh connection.
<ToyKeeper>
If I understand correctly, it won't be compatible with GNOME 50, because 50 removes all the X11 code from GNOME. But you could probably run the X11 version of GNOME 49, or any other X11 window manager.
stacksize has quit [Ping timeout: 244 seconds]
<iconoclasthero>
ok, thanks. not sure I have a long-term solution and from what I've tried, x over ssh has not worked for me with wayland.
eric_engestrom has quit [Ping timeout: 265 seconds]
eric_engestrom has joined #wayback
<axtlos>
iconoclasthero: X11 forwarding should work, though I never got that to work in general so i cant really test it
<iconoclasthero>
ah yes, waypipe. at one point (probably years ago now) I'd tried to get x forwarding working with wayland and was not successful...probably substantially due to the decision to use X11 as it "just worked." it looks like somethings gotta give in the next year if i don't want to give up a current GNOME DE.
<axtlos>
i mean, X forwarding with wayland isnt going to work
<iconoclasthero>
"waypipe is a proxy for Wayland clients. It forwards Wayland messages and serializes changes to shared memory buffers over a single socket. This makes application forwarding similar to ssh -X feasible" ?
<iconoclasthero>
there are a few more things that I use over ssh, but primariliy I need it for musicbrainz picard.
<axtlos>
well yeah, but its not X forwarding
<axtlos>
wayland itself wasnt really designed for uses like this, so waypipe is a kinda hacky solution
<iconoclasthero>
then let me back up and state what I'm looking for... I want to be able to launch picard on my remote machine over ssh as I'm doing all of the other related file stuff and have it pop up on my local machine with at least the same responsiveness as I currently have.
<iconoclasthero>
I want to be able to launch picard on my remote machine over ssh from BASH in my PWD, etc., as I'm doing all the other related file stuff*
<axtlos>
right, so if you're using wayland then you want waypipe
<iconoclasthero>
:thumbsup:
<iconoclasthero>
frankly, it would be nice if there was some sort of server/client functionality to picard, but the developers have been pretty adamant that's not on the table.
<seroquel>
whitequark[cis]: is there a bridge to this channel from matrix I can join?
<whitequark[cis]>
#wayback:catircservices.org
<seroquel>
cool ty
<whitequark[cis]>
Ariadne: should this be in the topic? smth like "| bridged to #wayback:catircservices.org (matrix)"
sq[m] has joined #wayback
<sq[m]>
sweet
seroquel has quit [Quit: leaving]
<sq[m]>
Is this is the appservice or the heisenbridge? I've only ever setup the appservice one, this instance seems a little slow but I haven't ruled out that it's my own home server, though my homeserver is generally pretty fast
<sq[m]>
I don't have irc bridging setup for anything atm, this is a new setup
<whitequark[cis]>
it's matrix-appservice-irc
<whitequark[cis]>
you can watch the latency of the bridge itself by using the live updates on libera.catirclogs.org
<whitequark[cis]>
actually now that i've typed it i realize that you can't since nobody is on catircservices.org
<whitequark[cis]>
(in any case, i think m-a-i is bad software but i keep using it for double puppeting)
<whitequark[cis]>
if it's lagging it's doing so in a way i can't see on the dash
<sq[m]>
I like the presentation of matrix and what I can do with it but yeah getting it to work well has been a bitch, I feel like even with the 8GB of RAM that generally it never uses that it really wants about 24GB at times, like if I change my avatar image, the server will send that event to every home server it possibly can which will take forever and will amass a ton of federation events that will start filling up memory pretty fast
<whitequark[cis]>
i have like 1.5G/4G used on that machine, can't be memory
<sq[m]>
disk is a problem too the database, I have mine on an ssd atm. I ran a homeserver on ec2 for awhile too and I feel like the magnetic storage really killed the performance compared to provisioned iops
<ari[m]>
but that's on a rather beefy machine, with plenty of cpu and ram to spare
<sq[m]>
I have netcrave.chat running at home now on a BHyve VM on an optiplex 7040m (mini)
<sq[m]>
yeah yours is probably fine, I'd just guess its maybe a little disk bottle-necked, my homeserver is just crazy because I'm joined to a ridiculous number of channels
<whitequark[cis]>
i'm running on hetzner cax11, that has SSDs supposedly. there's barely any disk activity and the CPU never even spikes to 100%
<whitequark[cis]>
i think it's just matrix being matrix
<sq[m]>
tru 😂
<sq[m]>
its finicky thats for sure
<ari[m]>
mine's also bare metal, some asrock rack barebones i bought a few years ago and fitted with ryzen 5950x, 128G of ram, 2T of nvme, and 4×6T of spinning rust
<whitequark[cis]>
how much did that cost you
<ari[m]>
i had the 2T and spinning rust laying around from before, rest of that slightly under 10kpln (under 2400€)
<whitequark[cis]>
that's... surprisingly cheap
<whitequark[cis]>
i mean, i only wish i could drop 2400k€ on a bare metal server to mostly sit around, but still it's surprisingly cheap
<whitequark[cis]>
actually wait i misread and didn't realize you had all of the storage already
<whitequark[cis]>
nevermind that's not very cheap :D
<ari[m]>
the barebones server - 1U chassis, motherboard, psu, all the cabling nicely done and all that, ready to plug storage/cpu/ram - cost around 3kpln (~700€) at the time, but shortly after i bought it, the price rose to ~1000-1200€
<ari[m]>
cpu was another roughly 700€, and ram was another roughly 700eur
<ari[m]>
also, hosting it costs me… significantly cheaper per month than a box from hetzner would
<whitequark[cis]>
hetzner costs me under 15 euro a month, including a 99% idle staging box that i keep around just to test changes before rollout to production
<whitequark[cis]>
it'd take me like a decade to break even with your setup
<whitequark[cis]>
but really it's not even about that, just about being able to drop 2 months rent on something that isn't the thing i earn money with
<ari[m]>
my previous setup was a hetzner box was 8 cores and 64G of ram, and renting that cost more than twice what i pay per month for this
<ari[m]>
but then again, i do actually need the cpu and ram
<whitequark[cis]>
that must be a really big matrix homeserver,
<ari[m]>
(it's doing more than just hosting matrix)
<sq[m]>
I'm using hurricane electric's IPv6 tunnel broker and a VPS in a DC here in SEA
<sq[m]>
~11ms to the VPS
<whitequark[cis]>
i remember being excited about he.net ipv6 tunnels back when they were new
<sq[m]>
this zimaboard is the router and I have 19 different routing tables (it's a little complicated)
<sq[m]>
yeah it was cool back in the day apparently they used to give people bgp sessions
<sq[m]>
now days you can't even use cloudflare with it because its blocked
<sq[m]>
otherwise I wouldn't even need the other VPS
<sq[m]>
this is sort of a job I've made for myself is figuring out how to make creating a dc @ home practical / safe
caskd has quit [Ping timeout: 245 seconds]
caskd has joined #wayback
<f_>
Ariadne: oh yeah can the matrix channel be mentioned in the channel topic
<f_|cis>
@whitequark:matrix.org: I have some tips and tricks to make matrix-appservice-irc less annoying on the IRC side
<f_|cis>
if you want to hear 'em
<f_|cis>
firstly the [m] prefix is configurable and you may want to set it to something other than [m] because some other bridges may use it and it may lead to conflicts because of that
<f_|cis>
then, the multiline pastebinning thing I'd set it to 6 or 10
<f_|cis>
The reply-to template I would set it to something like `$NICK: $MESSAGE (re $OG)` or something like that
<f_>
anyway I'm not too fond of matrix as you saw :p I mainly keep it around for postmarketOS and some matrix-only stuff
<f_>
for the topic, I do have op to set it but I dare not touch the topic unless Ariadne approves of the change :p
f_[m][m][m] has joined #wayback
<f_[m][m][m]>
test test
<f_>
oh also one more thing, I recommend using the MxID for nickname instead of the displayname .. but it seems you also bridge discord to irc with it, so maybe not.
<whitequark[cis]>
f_: re [m]: intentionally unchanged from the default, people kinda know what [m] is, less so for other suffixes
<whitequark[cis]>
re pastebinning, what's the default, 3? i thought 3 is when libera starts to throttle you
<whitequark[cis]>
re reply-to, worth changing i think, the default template is kinda awful
<whitequark[cis]>
re mxid: yep, won't work well with discord
<f_>
19:10 <whitequark[cis]> re pastebinning, what's the default, 3? i thought 3 is when libera starts to throttle you
<f_>
it says 3 but it really means anything that's 3 lines or more will be pastebinned. So this means every single multiline message except ones that are exactly 2 lines long
<f_>
might be annoying
<f_>
19:10 <whitequark[cis]> f_: re [m]: intentionally unchanged from the default, people kinda know what [m] is, less so for other suffixes
<f_>
alrighty
<whitequark[cis]>
mm yeah switching to 4 is a minimum then
<f_>
I like 6 or 7 TBH
<f_>
the pmOS bridge has it set to like 10 and had no issues whatsoever, but that's on OFTC
<whitequark[cis]>
i find it fairly annoying on support channels, since it's like only ever tripped by people new to irc etiquette
<whitequark[cis]>
what i'm thinking about here is the other direction, people pasting stuff into irc
<whitequark[cis]>
if i could set this per-channel i would be happy to set a looser limit here
<f_>
Regarding reply-to, matrixHandler:shortReplyTemplate, matrixHandler:longReplyTemplate and matrixHandler:longReplyTemplate, as well as matrixHandler:replySourceMaxLength
<f_>
whitequark[cis]: and btw you'll need to reboot the entire thing if changing these options.
<whitequark[cis]>
uhm.. want to send a PR? ^^
<whitequark[cis]>
you clearly know your way around this thing and i mostly do not
<f_>
I'll try
<f_>
but I'm currently backing things up on my laptop to move distros... so it will take a few days
<whitequark[cis]>
no worries
<f_>
but always happy to help
<whitequark[cis]>
i just dread touching the matrix stuff because there's always some gotcha and the next thing you know, 8 hours have passed and you're deep in the bowels of some nodejs package you didn't even know existed
<f_>
at least my irssi doesn't have nodejs :p
<whitequark[cis]>
it's not even the fact that it's a nodejs package that's the problem. it's that there are so fucking many of them
<f_>
I get you
<whitequark[cis]>
i write javascript code. my javascript code has an average dependency tree depth of, like, 2.
<f_>
when I tried to improve matrix-appservice-irc just a little bit I used docker to ease the task of getting the thing running locally in the first place
<whitequark[cis]>
i have a staging server that i'm free to break and such
<whitequark[cis]>
i assure you that using docker does not bring me any more joy than digging through suspiciously deep nodejs dependency trees
<f_>
whitequark[cis]: and I think you're good at it, else your light and working irc logger would've been a bulky and non-functional mess :p
<whitequark[cis]>
oh yeah
<whitequark[cis]>
back then i was actually pretty bad at frontend development. but i think i recognized that and, i guess, worked with that knowledge?
<whitequark[cis]>
if you look at the code it's kind of garbage. but it mostly works and there's little enough of it it doesn't really cause issues
<f_>
I mean
<f_>
as long as it doesn't take like 1 GB of memory just to view logs I'm happy
<f_>
(1 GB is roughly the amount of RAM element takes on my other f_[m][m][m] account)
<whitequark[cis]>
sure doesn't. all it does is lets you to link to messages/ranges (with graceful degradation) and use long polling (websockets didn't exist yet) to append messages to the log
<whitequark[cis]>
ok there's also two buttons that change a css class. you get the idea
<whitequark[cis]>
oh, and the clock is updated in realtime
<whitequark[cis]>
the clock is actually very funny. it shells out to cal on the server side to generate it. and then it parses it with regexps and highlights days as links. and then it updates the clock on the client side
<whitequark[cis]>
that's probably some of the worst code i've ever written
<whitequark[cis]>
the whole thing is so janky and held together with hopes and dreams and multiple pieces of long-deprecated software that every once in a while i kinda want to rewrite it. not fundamentally, just like... use websockets instead of long polling, yknow, that kind of stuff
<whitequark[cis]>
maybe tidy up the database schema
<whitequark[cis]>
and then i decide that eh, it'll work for another decade probably