adelaide has quit [Remote host closed the connection]
novakane has joined #river
snakedye has joined #river
<ifreund>
I think I'll move the repo from my personal acount to a github organization today, seems likely that I'll want to do that at some point in the future if the project continues to grow at its current rate
<ifreund>
any objections to "riverwm"?
<ifreund>
river-compositor is too long and plain "river" is already taken
<novakane>
make sense, sounds good to me
<novakane>
"wm" suffix is something everybody knows and recognises, best choice imo
<novakane>
also "riverwm" is way better to google :P
<ifreund>
I get "river wealth management" (riverwm.com) from DDG :D
<novakane>
oh yeah same with qwant lol
<novakane>
well that still less results than "river" :P
<dnkl>
riverwm sounds good
<ifreund>
done, now just need to update all the links I have control over
<ifreund>
though github will keep the old URL around for quite a while as a redirect in my experience
<novakane>
nice, I'm gonna update my url now before I forget
<novakane>
wlroots change, river change, what a day :P
<ifreund>
migration day all around :D
<novakane>
damn I have nothing to migrate, I want to participate :P
<novakane>
ifreund: you're gonna move zig-wayland/pixman and all in riverwm?
<novakane>
also I know we already talked about this but with an org maybe your view changed, still not interested in a river-protocols repo?
leon-p has joined #river
<leon-p>
speaking of being good to duck-duck-go, if you search for "river wayland", then you'll be surprised by just how many rivers and fishing supply shops there are in Wayland...
<novakane>
same problem when you search for awesomewm, I feel like wm require to be hard to search for :P
waleee has joined #river
<ecocode__>
I always loose my windows in all the tags. I soo much need riverctl to dump a list of windows :/
<leon-p>
a unique ID per window is definitely needed for `lswt | menu | riverctl focus` and similar pipes
<ecocode__>
It seems almost unbelievable river is working without some kind of pointer to each window
<ecocode__>
One day I will have to learn zig 😀
<ecocode__>
Probably not soon, cos I have raku and react already in the pipe ..
<ifreund>
novakane: no, splitting the protocols out into a river-protocols repo wouldn't make sense, I only intend for these protocols to have one server side implementation in river
<ifreund>
if multiple server side implementations were planed, they'd belong in wayland-protocols
<novakane>
fair enough
<ifreund>
leon-p: I think the issue you're seeing with resizing firefox and other slow clients on your PR is a likely a bug in the transaction which has been hidden till now by the fact that the x/y of views are frozen during resize
<ifreund>
and I don't think touching the transaction code right before a release is the best idea...
<leon-p>
ecocode__: Not sure what you mean by "pointer to each window", but the problem is not that river doesn't know its windows, it's that there is no way two uniquely identify the same window from multiple different wayland connections
<leon-p>
ifreund: agreed
<ifreund>
I really want to get that resize stuff merged because it feels way more intuitive for fast clients like foot, perhaps we can do a pretty quick second release with that feature and any necessary bug fixes
<leon-p>
that sounds good. If your blog posts gets submitted to link aggregator type sites, there will likely be an increase of bug reports for a month or two anyway
<novakane>
tbf it's more a performance problem than a bug, at least from a user point of view
<leon-p>
novakane: you can work around a client not having submitted a "good" buffer though
<ecocode__>
leon-p: I must be missing something... In sway, get-tree shows a json list of titles and id's. Is something like that doable ?
<novakane>
I mean at the end it resize the view like you want, with weird animations but still
<leon-p>
ecocode__: It's totally doable, trivial even. Although in river we'd probably just want a protocol extension for that instead of some json based IPC
<ecocode__>
👍
<leon-p>
ecocode__: The problem is, that although multiple wayland clients can query a list of all windows, there is no way for one of them to tell the other "I mean this one exact window".
<ecocode__>
Hmmm
<leon-p>
I wonder if unique IDs would make sense for the foreign toplevel protocol, or whatever version of it eventually makes it into wayland-protocols
<waleee>
the repo in the topic is still ifreund/river (which of course points to riverwm/river) but perhaps a topic update?
<ifreund>
Last todo before the release is to clean up my blog's code a bit so it's reasonable on mobile and whatnot
<leon-p>
ifreund: petition to rename your blog to "ifreuns iframe"
<ifreund>
:D
<novakane>
you finished your article?
<ifreund>
yeah it's ready
<novakane>
nice!
<ifreund>
I'm going to be busy with other stuff from now till pretty late this evening though I think, might not get it tagged today
<leon-p>
if you work on your blog, you might want to check for the new CSS media querry for dark theme preference. If you don't explicitly set it, I think firefox uses defaults which can look a bit bad
<ifreund>
hmm, yeah should probably do that
<leon-p>
also note that real phones work differently than the phone-imitation modes of both chrome and firefox, so don't expect it to actually work on real hardware. Guess how I know that.
* ifreund
is not much of a web dev
<novakane>
you use waybar so you're like a pro at css :P
leon-p_ has joined #river
leon-p has quit [Ping timeout: 268 seconds]
waleee has quit [Ping timeout: 268 seconds]
snakedye has quit [Ping timeout: 268 seconds]
snakedye has joined #river
leon-p_ is now known as leon-p
<doaN[m]>
did anyone tried River with Nvidia 495.44 ?
<doaN[m]>
it now supports GBM
snakedye has quit [Read error: Connection reset by peer]
snakedye has joined #river
n8_ has joined #river
<ifreund>
doaN[m]: river doesn't do anything differently than sway there, so it probably works decently
<ifreund>
properitary drivers are not and never will be officially supported by river or wlroots though
<doaN[m]>
yeah I understand
<doaN[m]>
I'll try it later.
<doaN[m]>
Thanks for making River ifreund
n8_ has quit [Ping timeout: 268 seconds]
<doaN[m]>
is there any plan to have rules for windows with tags and such ?
<doaN[m]>
I saw #456 but there isn't a clear answer.
<doaN[m]>
s/I saw #456 but there isn't a clear answer./I saw #456 but there isn't a clear answer/
snakedye has quit [Read error: Connection reset by peer]
snakedye has joined #river
snakedye has quit [Read error: Connection reset by peer]
snakedye has joined #river
snakedye has quit [Read error: Connection reset by peer]
snakedye has joined #river
anb has quit [Ping timeout: 264 seconds]
<ecocode__>
What hardware are you using? Laptops without Nvidia gpu are pretty uncommon to find.
<snakedye>
I think it's the opposite
<leon-p>
the average laptop does not have discrete graphics, luckily
<snakedye>
Indeed
<snakedye>
It's also a smoother experience in general
<ecocode__>
Well, I should have specified"high end" laptops😁
<leon-p>
putting a discrete graphics chip into your laptop as a manufacturer means way higher board design costs, way higher parts costs, higher material cost for beefier cooling, more space needed for the beefier cooling and needing a power distribution system capable of outputting more (often both voltage and amperes) which also limits the battery you can put in.
<leon-p>
considering that the average user simply does not care about having a discrete GPU, it would be a massive waste. And since it can significantly decrease battery runtime, even if the chip is not actively used, it can be a negative for a lot of people. I will most definitely never get a laptop with discrete graphics.
<leon-p>
same is true for memory higher than 8 GB, btw. Some people seriously claim that nowadays you need at least 16 gigs and some laptops already come with 32, but that is completely out-of-touch with reality
<novakane>
same for phone, when I saw the ram for an average one and I see the average user, totally useless
<ecocode__>
My 5yo machine has a dgpu and 32gb ram 🤪. I ordered a new one but delivery is for 2022 February. I happen to need a windows VM with 8gb ram, and doing pretty GPU intensive CAD. But I do agree with your comments
<leon-p>
ecocode__: It generally depends on what you classify as "high end". Sure, some manufacturers try to trick the average person into believing their products are good by sticking discrete graphics into them. But when you go up high enough that "consumer-grade" no longer applies, then manufacturers tend to know their target audience. For a lot of people battery runtime (meaning overall efficiency) and a small and light-weight
<leon-p>
body is considerably more important than having a discrete graphics chip.
<leon-p>
I consider every modern laptop with a battery runtime below 10 hours to be a bad joke
<novakane>
so every laptop then?
<leon-p>
hehe :P
<ecocode__>
The reason I ordered a new machine is that my GPU only has 2gb ram which I always reach badly. So I went for 8gb
<ecocode__>
If I get get 3 hours battery I will be more than happy!
<novakane>
damn
<ecocode__>
I need a portable machine to work at home or office. Both places I just close the lid and connect to external setup
<novakane>
well I'm good with 30min now since my battery is pretty dead :P
<novakane>
luckily I don't have windows on my laptop :D
<ecocode__>
Wow, did I just send a url to an emoji ?!
<novakane>
yeah lol
<ecocode__>
Bad phone, really bad phone!
<leon-p>
you have reached peak internet :P
<novakane>
I thought I was like rick rolled with an emoji :P
<ecocode__>
🤣
<ecocode__>
That worked !
<leon-p>
at least it's not a 500MB reaction gif that of course auto loads, luckily we are not /that/ platform (you which one I mean)
<leon-p>
s/you/you know/
<ecocode__>
At least I don't need the GPU in wayland applications, the CAD application still uses xwayland
<ecocode__>
And YouTube is ok with the igpu
<leon-p>
pretty sure hardware acceleration for Xwayland is a thing. Especially since every remotely modern (meaning from the last 20-ish years) somewhat complex X application does the same as on Wayland: client side rendering and pushing bitmaps to the server.
<doaN[m]>
<ecocode__> "What hardware are you using..." <- late reply but I have a desktop with a 980GTX
<doaN[m]>
s/980GTX/GTX 980/
<doaN[m]>
since I know 495.44 was coming with GBM support, I started for the first time a few days ago to use wayland with its ecosystem on a laptop with Nouveau.
<doaN[m]>
I must say River made me want to try it. And it is great.
<ecocode__>
@doaN[m] yep! river is like awesomewm for wayland, although it misses a lot of features, but it is on the right path. I tried sway more than once because it has all the possible features, but manual tiling is such a mess :(
<doaN[m]>
same didn't want to try sway because of its manual tiling
<doaN[m]>
* manual tiling but great project nonetheless.
<ecocode__>
👍
<leon-p>
you can probably have manual tiling without requiring a lot of mental load and implicit state. maybe limit the layout to a global row of window columns (maybe with automatic distribution) instead of a full tree.
<ecocode__>
is it possible to have rivertile set as layout manager for one tag and stacktile for another tag ?
<leon-p>
ecocode__: no, it's not. Layout generators can only be selected per-output
<leon-p>
when I came up with the first version of the protocol I did initially intend for users to run different layout generators side-by-side. However it seems like people generally prefer to use a single layout generator that is flexible enough to suit all their needs, so per-tag layout generators were never a priority
<leon-p>
FWIW you do not need to run stacktile and rivertile side-by-side because stackstile can generate all the layouts rivertile can as well
<snakedye>
leon-p_: I thought 16 gb was enough until I started using rust-analyzer 😅
<ecocode__>
@leon-p so I replaced rivertile by stacktile in my init file. this works, but sending commands i.e. ```riverctl send-layout-cmd stacktile primary_position right``` gives an error ```error: too many arguments```
<leon-p>
ecocode__: because you need to quote
<ecocode__>
euh ?
<leon-p>
riverctl only accepts a single argument as command for the layout generator
<leon-p>
if you have more, it will not concatenate them
<leon-p>
just note that stacktiles interface is not yet stabilized. I will change the format of commands significantly in the near future, mostly to allow sending multiple at once.
<ecocode__>
can I have one output using rows and another using columns ?
<snakedye>
Yes
<snakedye>
leon-p_: being able to run multiple layout generators at the same time is useful for debugging :)
<leon-p>
true
<ecocode__>
how can I specify to which output the riverctl send-layout-cmd should be applied ?
<ecocode__>
I mean in the init river file
<leon-p>
ecocode__: sadly, you can't, it will always go to the focused output. We have plans to change that, but it'll require some planning
<leon-p>
but
<leon-p>
but layout generators can have a different way (like comand flags) to set their defaults
waleee has joined #river
<snakedye>
bundling all the configuration in a single command is a little too much
<ecocode__>
ok. do layout generators get info about the windows (class, title,..) or do they only get the number of windows ?
<leon-p>
snakedye: nothing stops you from having a config file
<waleee>
hm and gnome-terminal in general it seems
<leon-p>
tl;dr: no. They used to in one version, but nobody used it and it had some issues, so it was removed for now. Might come back in some way, might not.
<snakedye>
True
<leon-p>
fun fact: wanting window app-id and title was the reason I brought up advanced layout generators in the first place. Before being wayland clients, they were just one-shot programs river called that worked by dumping a layout to stdout.
<snakedye>
One thing I could do if it was available now would be decorations for tabbed windows
<snakedye>
With the app-id
snakedye has quit [Ping timeout: 268 seconds]
snakedye has joined #river
snakedye has quit [Read error: Connection reset by peer]