| josch | i just connected my bluetooth headphones to the ax210 for the first time and discovered that apparently it's possible for one pair of headphones to be connected to two devices at the same time o0 | 00:01 |
|---|---|---|
| rick_ | josch this feature is quite old :D | 00:05 |
| josch | i also only joined this whole usb-c hype with the reform, leave me be with my ancient hardware :P | 00:05 |
| rick_ | for me bt died once too, but i didn't make much out of it cause a few hours later i had the same on my other linux laptop.. think maybe my new headphones are wonky | 00:06 |
| rick_ | did work fine with the old ones | 00:06 |
| josch | to reproduce chorc's issue i'm now going to watch a 45 minute episode on the pocket -- maybe this happens for me too | 00:06 |
| josch | (i want to see how common this is because bluetooth was flawless with a311d) | 00:07 |
| josch | fwiw, bluetooth survived but range is terrible -- i have to try out different antennas | 01:03 |
| minute | mhm | 01:09 |
| minute | i'm now in feierabend @ home and have 3 reforms running around me... classic, pocket and next :D | 01:10 |
| - erle (QUIT: Ping timeout: 260 seconds) (~erle@user/erle) | 01:11 | |
| josch | living the high life :D | 01:12 |
| minute | for me on classic reform i can kill the ax210 wifi (complete stall) when doing a big transfer over sshfs from another machine here | 01:15 |
| minute | since a few days... | 01:15 |
| minute | it will come back after a few minutes | 01:15 |
| minute | but there are no errors whatsoever | 01:15 |
| minute | yeah now it's back up... so weird | 01:16 |
| josch | only on classic? I just transferred a few gigs over ax210 on the pocket | 01:18 |
| minute | josch: yeah | 01:19 |
| josch | do you use it with an adapter in the minipcie slot without bluetooth? | 01:20 |
| minute | yeah | 01:29 |
| minute | mhm, there's newer firmware in apt | 01:29 |
| minute | wow, i'm getting 1055 mbit/s downstream on speedtest.net over wifi :0 | 01:33 |
| minute | a bit more than i pay for... | 01:33 |
| minute | hm, the sshfs issue isn't happening | 01:34 |
| minute | now | 01:34 |
| minute | but i also added a bunch of module parameters | 01:34 |
| minute | options iwlwifi power_save=0 bt_coex_active=0 swcrypto=1 uapsd_disable=0 | 01:36 |
| minute | options iwlmvm power_scheme=1 | 01:36 |
| minute | but the driver says these are "forbidden" so i'm gonna remove them > bt_coex_active=0 swcrypto=1 | 01:37 |
| - n_to (QUIT: Quit: quitidiquit) (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1) | 02:00 | |
| + n_to (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1) | 02:00 | |
| + paperManu_ (~paperManu@107.159.15.124) | 02:02 | |
| - chomwitt (QUIT: Ping timeout: 256 seconds) (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1) | 02:31 | |
| - mjw (QUIT: Ping timeout: 264 seconds) (~mjw@gnu.wildebeest.org) | 02:41 | |
| - paperManu (QUIT: Ping timeout: 240 seconds) (~paperManu@107.159.15.124) | 03:46 | |
| - pomel0 (QUIT: Ping timeout: 250 seconds) (~pomel0@user/pomel0) | 04:09 | |
| + pomel0 (~pomel0@user/pomel0) | 04:10 | |
| - pomel0 (QUIT: Ping timeout: 240 seconds) (~pomel0@user/pomel0) | 04:14 | |
| + pomel0 (~pomel0@user/pomel0) | 04:14 | |
| - bkeys (QUIT: Quit: With every step we take, danger will follow closely) (~Thunderbi@h193.131.19.98.dynamic.ip.windstream.net) | 04:45 | |
| - paperManu_ (QUIT: Ping timeout: 240 seconds) (~paperManu@107.159.15.124) | 05:08 | |
| + TechnoWizard (~TechnoWiz@user/TechnoWizard) | 05:28 | |
| - TechnoWizard (QUIT: Remote host closed the connection) (~TechnoWiz@user/TechnoWizard) | 05:28 | |
| - plomlompom (QUIT: Remote host closed the connection) (~plom@139.59.138.73) | 07:32 | |
| + plomlompom (~plom@139.59.138.73) | 07:33 | |
| + chomwitt (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1) | 07:53 | |
| + TechnoWizard (~TechnoWiz@user/TechnoWizard) | 08:03 | |
| - TechnoWizard (QUIT: Remote host closed the connection) (~TechnoWiz@user/TechnoWizard) | 08:03 | |
| - mlarkin (QUIT: Ping timeout: 244 seconds) (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 09:14 | |
| + mlarkin (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 09:16 | |
| - mlarkin (QUIT: Client Quit) (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 09:16 | |
| + mlarkin (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 09:16 | |
| grimmware | minute: have you seen the solderparty keebdeck? | 09:31 |
| grimmware | https://www.lectronz.com/stores/solderparty/ | 09:32 |
| + gustav25 (~gustav@c-78-82-54-254.bbcust.telenor.se) | 10:02 | |
| - mlarkin (QUIT: Quit: leaving) (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 10:20 | |
| + mlarkin (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 10:21 | |
| + nsc (~nicolas@i5C74DFDB.versanet.de) | 10:57 | |
| ch | Some devices (like the AX210 and 7265) with wireless and Bluetooth coexisting might have intermittent reliability with the iwlwifi driver. The development team suggests deactivating the bluetooth module. | 11:03 |
| ch | you'd think if the hardware has these features, they'd also work | 11:04 |
| josch | sometimes i feel like the FOSS community puts much more effort into ironing out bugs than the proprietary counterparts or at least is much more sensitive to bugs instead of ignoring them or working around the bugs. The Reform has issues but somehow they bug me more in the Reform than they bug me in some proprietary plastic computer. That's good because that motivates me to fix them. Just curious how I | 11:08 |
| josch | sometimes feel we apply different standards. | 11:08 |
| Asmadeus | Could just be that you can actually fix it for everyone and not just for you. I know I'm much more likely to let something bother me slightly if I know I can't possibly make the vendor fix it... | 11:10 |
| josch | right, i mean where would i even file a bug report with intel about their AX210? :/ | 11:11 |
| Asmadeus | We've got a problem at work with windows.. the i.MX6ull chip expects u-boot at the 1k offset for sd boot, so we make install disks with the gpt partition table carefully positionned out of the way, and apparently for some reason when you plug such a card into a recent windows they just recreate the partition table in the default location right over u-boot | 11:13 |
| Asmadeus | This literally took me a month to figure out how to report it in the first place (they apparently have a "feedback hub" app, couldn't find anything else), and now it's done it's just been ignored for a couple of weeks, it's great. | 11:14 |
| josch | windows just overwrites some bytes that it does not understand on the sd-card you plug in without asking the user first? Wow that is so bad o0 | 11:15 |
| Asmadeus | Yeah.. They also freely change the indices, even if the table is in the default position, I had disks with p1 and p10 (with some careful meaning about p10) be relabelled p1 and p2 without any interaction.. This obviously also happens when windows re-reads the parition table immediately after flashing the disk with rufus or similar, so the only workaround we found was to make our customers flash | 11:18 |
| Asmadeus | images from a VM | 11:18 |
| - jogu_ (QUIT: Remote host closed the connection) (~jogu@user/jogu) | 11:36 | |
| + jogu (~jogu@user/jogu) | 11:41 | |
| f_ | I cannot stand windows, you have my respect if you can | 12:16 |
| josch | ch: I saw you saw my pocket cable post. It was your message about the cable mess which made me think about this problem. I happen to have *two* of these acrylic inlays but only one Pocket. Would you like to have the second one? (also as a thank you for your last mail to me) | 12:22 |
| + mjw (~mjw@gnu.wildebeest.org) | 12:38 | |
| - MartiniMoe (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@user/MartiniMoe) | 12:38 | |
| + MartiniMoe (~quassel@user/MartiniMoe) | 12:38 | |
| - RandyK (QUIT: Remote host closed the connection) (~RandyK@user/randyk) | 12:58 | |
| + RandyK (~RandyK@user/randyk) | 12:58 | |
| ch | josch: ah. if you have a spare one i'll gladly take it. i'm | 13:14 |
| ch | opening the pocket way too often (: | 13:15 |
| josch | exactly my use case :) | 13:16 |
| josch | i'll mail it to the address that you put as the return address ony our package to me | 13:16 |
| + paperManu (~paperManu@107.159.15.124) | 13:21 | |
| ch | great, thanks | 13:49 |
| [tj] | I can’t get more than 640x480 out of the hdmi port on my imx8mp picket | 13:55 |
| [tj] | 13:55 | |
| [tj] | lol with a reboot it’s come up as 720x400 | 14:04 |
| - mrbcmorris_ (QUIT: Quit: ZNC 1.10.1 - https://znc.in) (~mrbcmorri@1513413-static.lxtnkya3.metronetinc.net) | 14:05 | |
| + mrbcmorris (~mrbcmorri@1513413-static.lxtnkya3.metronetinc.net) | 14:10 | |
| minute | [tj]: edid issue | 14:18 |
| + RandyK_ (~RandyK@user/randyk) | 14:29 | |
| - RandyK (QUIT: Ping timeout: 272 seconds) (~RandyK@user/randyk) | 14:29 | |
| * RandyK_ -> RandyK | 14:30 | |
| ch | josch: it occurs to me you could just hand it to me at 39c3 | 14:37 |
| ch | josch: spare the transport etc | 14:38 |
| - jn (QUIT: Ping timeout: 244 seconds) (~quassel@user/jn/x-3390946) | 14:47 | |
| + jn (~quassel@2a0a-a54a-a6d6-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de) | 14:48 | |
| - jn (QUIT: Changing host) (~quassel@2a0a-a54a-a6d6-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de) | 14:48 | |
| + jn (~quassel@user/jn/x-3390946) | 14:48 | |
| [tj] | minute: is there a work around? | 14:53 |
| minute | [tj]: sway or gnome? | 14:53 |
| [tj] | I’ll plug in directly rather than through the kvm, but this is the first device to ever have an issue | 14:53 |
| [tj] | Wayfarers | 14:53 |
| [tj] | Wayfire | 14:53 |
| minute | oh, ye olde | 14:53 |
| minute | [tj]: imx? | 14:54 |
| [tj] | Yeah | 14:54 |
| minute | hmhm. but the good news, wayfire can set a custom modeline | 14:54 |
| minute | like sway | 14:54 |
| minute | [tj]: random googling https://forums.raspberrypi.com/viewtopic.php?p=2275350&sid=bb20baf3ac7c09fca8a1f36fe34454d1#p2275350 | 14:56 |
| minute | [tj]: https://github.com/WayfireWM/wayfire/wiki/Configuration#outputcustom_mode_n | 14:57 |
| [tj] | Setting the resolution from wayfire config managers doesn’t work, aren’t these just setting config? | 14:59 |
| [tj] | wlr-randr let me at least see the modes which are valid and pick one of those | 15:04 |
| minute | [tj]: i have no experience with config managers | 15:06 |
| minute | [tj]: so if wlr randr shows you available modes, edid works or no? | 15:07 |
| minute | [tj]: also, i'm not talking about selecting a resolution. but a modeline. | 15:07 |
| minute | i.e. you give the system all parameters of the mode, so it doesn't need to look at edid | 15:08 |
| minute | anyway, i think i've given you all required information to set up the mode you need. now afk to take care of life things | 15:09 |
| + erle (~erle@user/erle) | 15:31 | |
| - jn (QUIT: Ping timeout: 244 seconds) (~quassel@user/jn/x-3390946) | 15:43 | |
| + jn (~quassel@user/jn/x-3390946) | 15:44 | |
| - Guest3190 (QUIT: Ping timeout: 265 seconds) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 16:02 | |
| - GNUmoon2 (QUIT: Ping timeout: 272 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 16:13 | |
| + Guest3190 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 16:15 | |
| + GNUmoon2 (~GNUmoon@gateway/tor-sasl/gnumoon) | 16:37 | |
| josch | [tj]: another thing to try: flash an sd-card with the latest system image and start gnome and see what happens then. Maybe it's "just" a software issue? | 16:41 |
| ch | minute: https://shop.mntre.com/products/mnt-pocket-reform should have an image of https://shop.mntre.com/spree/products/432/large/mnt-reform-rcore-rk3588-processor-module-edit.jpg?1709746051 but that 404s | 16:42 |
| minute | ch: ah dang, thanks! | 16:53 |
| josch | minute: maybe on that page it should also not say "PCIe 2.0 x1 on M.2 key M slot." but mention the additional M.2 slot on the new headphone switch board? | 16:53 |
| josch | (it does mention it under "Networking") | 16:54 |
| minute | josch: ch: would you mind filing that on mnt.re/bugs ? | 16:54 |
| josch | oh no it does not | 16:54 |
| josch | sure | 16:54 |
| josch | done :) | 16:57 |
| josch | dang ch was faster | 16:58 |
| josch | i blame that typing in gitlab is laggy as hell | 16:58 |
| josch | (letters appear in the textbox at about 2fps or so) | 16:58 |
| minute | waa | 16:59 |
| minute | on a311d + firefox? | 17:00 |
| minute | i'd like to see a flamegraph of that... oof | 17:00 |
| josch | You'd like to see that? What would be an easy way to profile it? | 17:02 |
| erle | could anyone quickly try out if oneko works in xwayland for you? it works for me on an intel GPU, so i think it may just be a driver bug instead of an xwayland bug. | 17:02 |
| josch | (yes i'm still on a311d and yes i know...) | 17:02 |
| josch | erle: the cat only moves when i put my cursor on top of it but not otherwise | 17:04 |
| minute | that's a feature | 17:04 |
| josch | and the reason why it's harder to automate things :( | 17:05 |
| josch | with wayland, oneko can no longer arbitrarily query what my input devices are doing (security feature) | 17:05 |
| minute | josch: what do you need to automate that's hard? | 17:07 |
| minute | also, you can always use Xephyr | 17:07 |
| josch | minute: when i want to automate, for example, reform-setup-wizard i need to know where each button is rendered to decide where to automate the click. With wayland, another client can no longer find out these things about other clients. Gnome accessibility works around this by having a daemon running at all times which keeps track of widget positions. | 17:10 |
| josch | i used to use dogtail for that under xorg. Under wayland it's a bit more complicated: https://modehnal.github.io/ | 17:12 |
| josch | and since this needs a gnome session i cannot easily automate the reform-setup-wizard as run under sway | 17:12 |
| josch | i currently hardcode button positions | 17:12 |
| josch | ah right the gnome component was called ponytail | 17:13 |
| minute | josch: ah. i wonder how orca works then | 17:27 |
| minute | josch: also, you could test in Xephyr using ye olde X tools, gtk still runs on X, right? | 17:28 |
| josch | minute: yes, but i'd like to test in the same environment that things are run in. And yes, i also wonder about orca. | 17:33 |
| josch | minute: here is a profile session from firefox while i try to type a long text into the MNT gitlab issue form: https://share.firefox.dev/3XdlI6S | 17:33 |
| josch | there is a flame graph but it doesn't tell me anything | 17:34 |
| + paperManu_ (~paperManu@142.169.16.166) | 17:36 | |
| josch | that looks like a lot of minified javascript | 17:36 |
| minute | josch: the bars in the top are the interesting stuff i think | 17:37 |
| minute | josch: there are some red bars, when hovering, it says "jank" | 17:37 |
| minute | josch: https://minute.social/secret-room/Screenshot%20From%202025-11-22%2017-38-05.png | 17:38 |
| minute | josch: to my untrained eye it looks like most time is spent in garbage collection?! | 17:40 |
| minute | josch: on rk3588, typing in that gitlab text field is also quite laggy, but not that extreme | 17:44 |
| josch | uff if it's laggy on rk3588 then you can imagine the slide-show over here :D | 17:45 |
| minute | haha i tried it myself now | 17:46 |
| minute | profiling it | 17:46 |
| minute | looks like every keypress consumes like 1 MB of memory | 17:46 |
| josch | the simple workaround is to write the markdown in an external editor and then copypasting it into the browser tab | 17:46 |
| minute | what have we done to software... | 17:46 |
| josch | minute: you really have strange hobbies :) | 17:46 |
| ch | ouch | 17:50 |
| minute | i think they're using some really non performant frontend framework maybe | 17:52 |
| - paperManu_ (QUIT: Ping timeout: 246 seconds) (~paperManu@142.169.16.166) | 17:54 | |
| minute | uhm, the textarea has like 100 event handlers and they all remove the same css class? | 18:04 |
| minute | disabling those event handlers makes it fast | 18:06 |
| + kxtells (~kxtells@user/kxtells) | 18:17 | |
| kxtells | Hello! It's me, the keyboard problem guy :-) Just wanted to share my recent findings. | 18:17 |
| kxtells | I was running the pocket with two 7565121 batteries, and this morning I saw what seemed a brown out | 18:18 |
| kxtells | I switched to stock batteries, the brown out did not happen again | 18:18 |
| minute | they use vuejs _and_ jquery together | 18:18 |
| kxtells | and the keyboard, seems to work much better now. I can still see sporadic `reset full-speed` but the connection has not been fully lost in the whole day | 18:18 |
| minute | kxtells: hey, do you have reform-power-daemon already? if not, please try it | 18:18 |
| minute | kxtells: (should be just `apt update` and `apt install reform-tools` | 18:19 |
| kxtells | I will try it for sure. Was mostly surprised to see the unexpected goood behaviour of the keyboard | 18:19 |
| minute | kxtells: nice, what changed about the keyboard? | 18:19 |
| kxtells | @minute I barely get any `reset full-speed` messages, and so far the keyboard is still connected. I am running plugged today, I will check unplugged tomorrow | 18:21 |
| josch | kxtells: reform-power-daemon was added in the latest version of reform-tools. If you experience any issues, please let us know. :) There was a bug (should now be fixed) where after installing the latest reform-tools version, reform-power-daemon wasn't enabled and started. You can make sure that it is running by looking at sudo systemctl status reform-power-daemon.service | 18:22 |
| minute | kxtells: sorry i mean, what did you change to make it better? | 18:23 |
| kxtells | thanks | 18:23 |
| kxtells | @minute nothing, that is the funny part. I changed the batteries | 18:23 |
| minute | kxtells: ah ok. hm | 18:24 |
| minute | maybe we should migrate to forgejo next year... i don't like slow tools | 18:24 |
| josch | but where is my motivation to upgrade to rk3588 then??? :D | 18:25 |
| josch | (i run my own forgejo and only have made good experiences so far but i don't have CI set up, just basics) | 18:25 |
| minute | it's good that it's using a compiled language instead of rails also | 18:25 |
| minute | for gitlab we have to constantly do security updates | 18:26 |
| erle | forgejo seems significantly faster UI wise (i use it sometimes), but it also has some js hickups. and also ofc it does not have feature parity. its a bit of time since i used gitlab in a corporate way, but i supect that “require 1 person of group A and 1 person of group B to approve this MR” might not be a thing that forgejo can do (gitlab can though). | 18:26 |
| minute | ok, we don't have such highly specific workflow needs | 18:27 |
| josch | minute: did you happen to have a look at the guid partitioning MR? I think it can be merged. | 18:27 |
| erle | the js hickups in forgejo i have seen were not lag though, only “looking at the same page with and without js is the same except with js there is some background activity that prevents CPU from idling”, then again, that's negligible | 18:27 |
| ch | erle: iirc the community edition of gitlab also cannot do these approval requirements | 18:28 |
| erle | ch ah i see! | 18:28 |
| josch | erle: you have high standards. We are comparing it with: "typing in gitlab issue textbox happens at 1 fps" :D | 18:28 |
| f_ | it's not that bad right? | 18:28 |
| erle | josch too many websites having such bugs are the reason i use noscript almost everywhere :( | 18:28 |
| f_ | rrright?? | 18:28 |
| erle | f_ see what minute wrote: <minute> looks like every keypress consumes like 1 MB of memory // <minute> uhm, the textarea has like 100 event handlers and they all remove the same css class? | 18:29 |
| minute | erle: there doesn't seem to be a standard way to remove event handlers in DOM/JS without knowing the handler function somehow? otherwise i could make some bandaid patch | 18:30 |
| f_ | I use postmarketOS gitlab all the time and the textbox isn't particularly laggy.. | 18:31 |
| minute | f_: on what kind of machine? | 18:31 |
| josch | f_: are you using it from a a311d? :) | 18:31 |
| minute | also i could imagine it's probably faster in chromium (sigh) | 18:31 |
| f_ | minute, josch: that same ol' 2011 laptop (i7-2630qm), gitlab running in Tor Browser which is firefox-based | 18:32 |
| f_ | maybe they added more junk since pmOS gitlab's version | 18:32 |
| minute | > Largest Contentful Paint (LCP) > 3.58 s | 18:34 |
| josch | minute: it is faster in chromium but still not as fast as i type | 18:34 |
| minute | > Your local LCP value of 3.58 s needs improvement. | 18:34 |
| minute | lol | 18:34 |
| minute | (chromium "performance" tab) | 18:34 |
| f_ | but also (not scientific test) web browsing on S905D3 isn't all that fast | 18:34 |
| minute | > LCP element label.custom-control-label | 18:34 |
| f_ | (amlogic S905D3 came after A311D) | 18:34 |
| f_ | (and this is on my S905D3 board with .. can't remember what config I put in there) | 18:35 |
| minute | josch: ah, can you go to Performance tab in chromium dev tools and then continue typing in that field? | 18:35 |
| f_ | but I suspect it might be ram.. | 18:35 |
| minute | josch: the stats there are quite interesting | 18:35 |
| minute | josch: it gives delay in ms and you can open that and it shows input delay, processing duration and presentation delay | 18:35 |
| f_ | I frequently run out of ram on this workstation (671 MB free right now, out of 8 GB) | 18:36 |
| minute | josch: with this data we could file a bug at gitlab | 18:36 |
| f_ | Actually .. textbox *is* slow here .. but not extremely slow either, it feels a tiny bit laggy sometimes, but just a tiny bit | 18:37 |
| erle | minute IMO removeEventListener() is a very frustrating interface, given that it needs the same *reference* to the function used in addEventListener(). i think you might be able to use cloneNode() to clone an event without its listeners, like elem.replaceWith(elem.cloneNode(true)); https://developer.mozilla.org/en-US/docs/Web/API/Node/cloneNode | 18:37 |
| erle | > Cloning a node copies all of its attributes and their values, including intrinsic (inline) listeners. It does not copy event listeners added using addEventListener() or those assigned to element properties | 18:37 |
| josch | Your local LCP value of 6.11 s is poor. | 18:38 |
| minute | erle: ah interesting | 18:42 |
| erle | minute i don't have time right now, but if you use the clone trick, tell me if it works | 18:42 |
| minute | most time is spent in updateNewWorkItem | 18:42 |
| minute | which is part of some graphql client | 18:42 |
| minute | that has some strange "local in memory cache" implementation in javascript | 18:42 |
| josch | minute: i don't know what i'm doing but it looks like this: https://mister-muffin.de/p/UkrJ.png | 18:43 |
| minute | josch: 665ms input delay wow | 18:43 |
| minute | josch: if you zoom in in "main" you'll see function names | 18:44 |
| erle | if i had ultimate authority over a team of webdevs, i'd give them underpowered 2010-era netbooks and a throttled 3G connection. that will teach them how to make stuff perform ;) | 18:45 |
| minute | josch: this is a flame graph that shows the call traces in boxes and the longer the box is, the more time that call took | 18:45 |
| minute | josch: and boxes that are lower (on y axis) than boxes above them are nested function calls | 18:45 |
| minute | josch: it's like the gnome sysprof, just upside down | 18:45 |
| erle | “i think you might be able to use cloneNode() to clone an event without its listeners,” should have been “i think you might be able to use cloneNode() to clone A NODE without its listeners,” | 18:45 |
| erle | sorry | 18:45 |
| minute | erle: yeah maybe we should use that as a sales angle to webdev teams... here's a slow laptop to try out your bloat on | 18:46 |
| josch | ah nice i now see how i can "zoom in" | 18:49 |
| josch | ah now i see the vue.js calls -- this is quite a bit smoother to use than the similar view in firefox | 18:52 |
| minute | josch: what happens if you click on "rich text editor"? | 18:52 |
| minute | on the bottom of issue text entry. is that faster? | 18:53 |
| minute | it seems faster for me than the raw text input | 18:53 |
| josch | it is nearly smooth in firefox | 18:54 |
| josch | there is a lag between keybress and letters showing up but at least it manages to keep up with my typing speed | 18:56 |
| josch | *keypres | 18:56 |
| minute | the rich text editor? | 18:56 |
| josch | keypress | 18:56 |
| minute | so it's faster than the normal input? | 18:56 |
| josch | yes, rich text editor is noticibly faster | 18:56 |
| josch | yes | 18:56 |
| minute | strange :D | 18:56 |
| minute | but hey, at least a little relief | 18:56 |
| josch | it does not make sense that it's faster XD | 18:56 |
| minute | yeah it's a bug | 18:57 |
| minute | but i guess it wasn't noticed | 18:57 |
| minute | because the developer's machines are too fast | 18:57 |
| f_ | I bet the developers are all running windows 12 on amd threadrippers | 18:57 |
| minute | i'd think they run mac OS but not sure :D | 18:57 |
| f_ | maybe hackintosh :D | 18:58 |
| f_ | or that'd make sense: M1 MacBook they replace every 6 months | 18:59 |
| josch | minute: sorry for my unintended nerd snipe with this topic :) | 18:59 |
| minute | at least most of the commercial/closed source web developers i met in my career used macs | 18:59 |
| minute | josch: haha no problem. i was curious about the lags in gitlab for a while. | 18:59 |
| erle | it's a classic thing btw. you see it more with games and web stuff. | 19:20 |
| erle | minetest/luanti issue tracker and irc has some amazing discussions that reveals how much better hardware people programming the game have than users. “should we make the default viewing range 1000?” :P | 19:21 |
| erle | (it's funny how people with super fast computers often think users get a bad first experience if not all effects are set to max by default, but rarely think that lag is also bad first experience) | 19:23 |
| minute | old games from the 2000s usually did some system check and adjusted performance settings for your setup | 19:30 |
| minute | (i.e. Deus Ex) | 19:30 |
| f_ | some games still do that | 19:33 |
| minute | right. but luanti doesn't i think? | 19:33 |
| erle | minute the devs ripped out the opengl 1.x renderer, even though it worked better on hardware that supports opengl 2.x than the opengl 2.x renderer. they don't care about old or low-spec hardware, because they don't use it. | 19:41 |
| erle | benchmarking on your own setup only is always going to lead to such things. i remember when lizzy (former mineclone2 dev) once made a mistake that could result in megabytes of traffic when a player was on fire … she only tested it on LAN/loopback. | 19:43 |
| + S0rin (~S0rin@user/s0rin) | 19:43 | |
| minute | erle: why did they rip it out, was it a maintenance burden? | 19:45 |
| erle | also cora once wrote fire code where every node that was on fire had a timer … but fire “copies itself” by putting other things on fire. i caught the memory problem in review by settting a mineclonia forest on fire on a machine with 2GB RAM and waiting … then cora changed it so that fire extinguished randomly, because that means not having to track state for each fire node. | 19:45 |
| minute | i.e. i understand deprecating and removing stuff when you can no longer maintain it, otherwise if it bitrots, people will be upset that it doesn't work | 19:46 |
| erle | minute if you look into the git repo, they have a habit of ripping out functionality if they don't like it or don't use it since about minetest 5.3. it accelerated after 5.5. | 19:46 |
| minute | sorry, no time for that atm :D | 19:46 |
| minute | "less code is debugged code", right? | 19:47 |
| minute | s/less/deleted | 19:47 |
| erle | sometimes “old and rarely-changed and users get angry if you rip it out” means “battle-tested” | 19:47 |
| erle | i have offered to maintain the renderer in case someone thought it was a maintenance burden actually. | 19:48 |
| erle | minute long story short, at some point i got banned from the issue tracker for one year while working on stuff only affecting low-spec or old hardware, and *exactly* one year later they ripped out the renderer and at least one user who complained about the performance issues was asked if they are me. | 19:50 |
| erle | i have zero intention bothering them about support again, because i understand perfectly what message that was supposed to send. | 19:50 |
| erle | comments like “if someone would be opposed to this measure, they could complain on the bug tracker” coming from devs who ban people who do indeed complain are not exactly subtle. | 19:51 |
| erle | (and before anyone accuses me of being uncivil, all the chatlogs and issues are in the open. you can decide for yourself.) | 19:51 |
| erle | and i did contribute stuff, not just complain. anyway, if you want to re-add the renderer, it's probably a diff of <1500 lines unless something big has changed. | 19:52 |
| erle | i have to go | 19:53 |
| erle | oh, last thing: if luanti >5.10 is slow on reform, downgrade to 5.10 and disable shaders. | 19:56 |
| erle | mineclonia supports 5.10 (it's in debian stable), so that will work | 19:56 |
| erle | >5.10 will still work btw, but with noticeably worse performance. you can ofc always reduce the view distance to get good performance unless you are using mesa software rendering, but using the opengl 1.x renderer should make it possible to have a smooth game of mineclonia at view distance 100. | 19:57 |
| erle | which about matches the performance of a thinkpad from 2006 | 19:58 |
| erle | see ya | 19:58 |
| - erle (QUIT: Quit: K-lined) (~erle@user/erle) | 19:58 | |
| Zaba | it seems especially ironic given that opengl 3 is not exactly state of the art either these days | 19:59 |
| [tj] | josch: yeah I should probably update, it’s just that I hate updating computers | 19:59 |
| josch | [tj]: i'm not necessarily suggesting an upgrade. Just to isolate where the problem is. I have systems that still run Debian Stretch. I also hate upgrading. :) | 20:22 |
| - jn (QUIT: Ping timeout: 250 seconds) (~quassel@user/jn/x-3390946) | 21:35 | |
| + jn (~quassel@2a0a-a54a-2ff8-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de) | 21:35 | |
| - jn (QUIT: Changing host) (~quassel@2a0a-a54a-2ff8-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de) | 21:35 | |
| + jn (~quassel@user/jn/x-3390946) | 21:35 | |
| minute | experiments https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/153/diffs | 21:42 |
| minute | (potentially dangerous) | 21:42 |
| - gustav25 (QUIT: Quit: Quit) (~gustav@c-78-82-54-254.bbcust.telenor.se) | 22:15 | |
| - schalken (QUIT: Ping timeout: 264 seconds) (~schalken@117-118-178-69.gci.net) | 22:21 | |
| minute | ready to install, here goes nothing :33 | 22:22 |
| + schalken (~schalken@117-118-178-69.gci.net) | 22:23 | |
| minute | ch: for me, the red charging led is off at 100%, not sure if bug, but i think confusing. esp in no-battery case. see latest forum post. before, the red led was a nice "charger negotiated" feature. | 22:31 |
| minute | weird, neither of my new patches did anything. | 22:39 |
| minute | oh what's this https://github.com/JeffyCN/drm-cursor | 23:01 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
