| minute | haha | 00:04 |
|---|---|---|
| josch | if we split the package, then maybe put all the desktop customizations in one package | 00:04 |
| josch | like, everything in /etc/skel | 00:04 |
| josch | and then call it reform-desktop | 00:09 |
| + erle (~erle@2a02:3102:4286:16:be:bdbd:2409:d4a5) | 00:50 | |
| - mtm (QUIT: Ping timeout: 256 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 01:03 | |
| minute | josch: good idea | 01:40 |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 01:47 | |
| mhoye | I was just coming in here to ask if reform-tools could not depend on gnome-system-monitor | 01:57 |
| mhoye | but I see that's already in play | 01:57 |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 02:07 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 03:09 | |
| - mjw (QUIT: Ping timeout: 260 seconds) (~mjw@gnu.wildebeest.org) | 03:09 | |
| - jacobk (QUIT: Ping timeout: 268 seconds) (~quassel@utdpat241106.utdallas.edu) | 03:35 | |
| - nsc (QUIT: Ping timeout: 260 seconds) (~nicolas@182-49-142-46.pool.kielnet.net) | 03:44 | |
| + nsc (~nicolas@225-48-142-46.pool.kielnet.net) | 03:45 | |
| + jacobk (~quassel@utdpat242026.utdallas.edu) | 04:03 | |
| - erle (QUIT: Quit: Democracy must always be better armed than tyranny.) (~erle@2a02:3102:4286:16:be:bdbd:2409:d4a5) | 04:34 | |
| - jacobk (QUIT: Ping timeout: 268 seconds) (~quassel@utdpat242026.utdallas.edu) | 04:57 | |
| + jacobk (~quassel@utdpat242026.utdallas.edu) | 05:21 | |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@utdpat242026.utdallas.edu) | 06:16 | |
| + jacobk (~quassel@129.110.242.173) | 06:37 | |
| josch | i'd still prefer to first go the route of demoting dependencies to recommends by making them optional after changing the code itself | 07:26 |
| josch | because just splitting the package into two doesn't help at all immediately | 07:30 |
| josch | imagine we'd do the split, then reform-tools would still need to Depend on reform-desktop because if it does not, anybody updating reform-tools would *not* get reform-desktop installed, thus potentially breaking their setups | 07:31 |
| josch | mhoye: the case for gnome-system-monitor is the same as the one for thunar: libexec/reform-tools/reform-tray.py could be changed such that its dependency on gnome-system-monitor becomes optional | 07:32 |
| - hairu (QUIT: Remote host closed the connection) (m-uotkmd@user/hairu) | 08:02 | |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@129.110.242.173) | 08:19 | |
| - Gooberpatrol66 (QUIT: Ping timeout: 264 seconds) (~Gooberpat@user/gooberpatrol66) | 08:31 | |
| + jacobk (~quassel@64.189.201.150) | 08:45 | |
| + Gooberpatrol66 (~Gooberpat@user/gooberpatrol66) | 09:22 | |
| hramrach | josch: I think it makes sense that somebody would not want the GUI part because of reform-tools they use something else that cannot make use of that tray icon - eg mlterm directly on console. While it's not a setup I would like I understand that different people use fifferent things, and the GUI is optional | 09:23 |
| + hairu (m-uotkmd@user/hairu) | 09:24 | |
| hramrach | ugh, grammar | 09:24 |
| hramrach | also splitting is problematic because unless you split into a lot of packages somebody will always complain that to get this optional part they also have to get that part they did not want | 09:28 |
| hramrach | it should be possible to do the split so that people who have the old package get all the new parts even if they are optional but not sure about the packaging details | 09:33 |
| hramrach | also if the part that goes mising truly is not critical then reinstalling it after the split is an annoyance but not really something that can be called broken 🤷 | 09:42 |
| josch | hramrach: i don't disagree with either splitting or making it optional but splitting the package takes much more time because we essentially have to wait for a sufficiently long time such that everybody upgraded to the new version before we can drop the dependency of reform-tools on reform-desktop | 10:20 |
| josch | so such a split will not help anybody today | 10:20 |
| josch | what should help everybody who wants a more minimal setup today is to demote desktop packages from Depends to Recommends | 10:20 |
| josch | and this should fix that problem: https://paste.debian.net/hidden/94662555/ | 10:21 |
| josch | and that then allows us to demote a bunch of stuff to Recommends: https://paste.debian.net/hidden/d82c0073/ | 10:29 |
| josch | and here is the diff for reform-system-image which explicitly installs the Recommends of reform-tools to make sure they are available in the default install: https://paste.debian.net/hidden/563323e5/ | 10:36 |
| hramrach | what always bothered me is that there is not a clear way to find recommended packages that are not installed | 10:37 |
| josch | hramrach: aptitude search '~RBrecommends:~i' | 10:38 |
| hramrach | Thanks, so there is a way, albeit not a clear one | 10:38 |
| josch | in our case, reform-check will tell you if the packages are missing | 10:38 |
| josch | hramrach: i think synaptic has a custom "missing recommends" filter if you like to click buttons | 10:39 |
| hramrach | why can't the ease of understanding of clicking buttons be transferred to the commandline tools? | 10:40 |
| hramrach | Surely i's not that whenever it's a commandline tool it must use cryptic syntax along the lines of sendmail.cf | 10:41 |
| minute | josch: yes, i thought about it some more. lets make the dependencies optional instead, and fiddle with reform-tray so it doesn't offer stuff that's not installed | 11:03 |
| minute | josch: what i want to say is i totally agree with what you said | 11:05 |
| josch | cool, thanks! :) | 11:41 |
| josch | hramrach: i agree, easy to use CLI interfaces for package management would be nice. I yet have to see one that ticks all the boxes though. I spent the last 13 years of my life developing utilities around the package management problem and there are still many tasksn for which i have to resort to my custom scripts because tools like apt do not offer the right interface. If you have a plan, I'd love to | 11:44 |
| josch | hear it. I don't mean this in a mocking way. I think I'm way too much specialized on the topic to be able to design such an interface myself. | 11:44 |
| josch | minute: i refactored reform-tray a bit to avoid too much code duplication: https://source.mnt.re/reform/reform-tools/-/merge_requests/65 | 11:54 |
| josch | i also pushed the required corresponding change to the reform-system-image MR: https://source.mnt.re/reform/reform-system-image/-/merge_requests/91 | 12:07 |
| hramrach | I thinks that's large part of the problem with many opensource projects - at the point anyone is able to contribute changes to the project they are so deep in they compeletely lose the perspective of somebody not familiar with all internals. There is very little UX research done for software developed commercially, and for opensource it's basically non-existent. | 12:07 |
| josch | since reform-tools 1.37 with support for pocket-reform-system-a311d.img, pocket-reform-system-imx8mp.img and reform-system-imx8mp.img is now in the repo, I'm going to merge this MR to main once it succeeds, okay? | 12:08 |
| minute | josch: this looks great, awesome refactoring + i like the idea of disabling the items the way you did it | 12:08 |
| josch | hramrach: +1 | 12:08 |
| josch | minute: what i don't like about it is that i failed to add a tooltip to the menu items | 12:09 |
| josch | minute: i would've liked to have a tooltip which shows *why* the item is grayed out when one hovers over it with the mouse | 12:09 |
| minute | my 2 cents for the package management topic: i think instead of those gnome and kde etc "app store" like UIs with ratings and whatnot, there really should be just a modern (lets say gtk4) frontend for apt, i.e. a reimplementation of aptitude | 12:10 |
| minute | josch: tooltips are hard in that context yeah | 12:10 |
| minute | sorry not aptitude, synaptic | 12:10 |
| minute | i am not fully awake yet | 12:10 |
| minute | coffee has not passed blood brain barrier | 12:11 |
| hramrach | Still there is this cursorless project that put some thought into makeing voice control interface manageable, and while I have not tried to use it so far I can appreciate that they came up with new paradigm to make things work with that intrface. So it's possible to do come up with new interface concept that fits a specific problem bu it's not trivial. | 12:12 |
| hramrach | I don't think making it graphical really helps | 12:14 |
| hramrach | What feature of synaptice depnds on graphics that cannot be provided in commandline? | 12:15 |
| hramrach | For context I have used synaptic like twice and since seeng it avoided it like the plague | 12:15 |
| hramrach | So there might be some geat feature I missed because the initial experience was so bad I gave up on the software right away. | 12:18 |
| minute | ok, we're probably talking about totally different things, sorry | 12:18 |
| minute | i personally don't know how to design ergonomic, discoverable command line tools with a large number of features | 12:19 |
| hramrach | looks like pretty much nobody doing commandlkine tools does | 12:20 |
| hramrach | one pretty ancient thing that comes to mind is that 'dumb terminal wizard' system where the options are prinet each one a line and numbered, you select an option by number | 12:21 |
| minute | fdisk comes to mind | 12:22 |
| hramrach | right, that's using pretty much that interface | 12:22 |
| minute | i see that this would be helpful for voice control. is that something you require? | 12:22 |
| minute | it's like a telephone computer interface | 12:22 |
| hramrach | no, I gave cursorless as an example of a project that came up with new interface paradigm for the problem they were tying to solve because existing ones did not work | 12:23 |
| minute | i know how to make GUI tools that help with discoverability and i think i have some visual design sense. i think synaptic as it is is not good, doesn't work properly on wayland without hacks etc, that's why i also don't use it. | 12:24 |
| hramrach | but there is also the thing that if accessibility tools relly worked people who do not really need them would use them when they feel lazy because it's that simple | 12:24 |
| minute | intuitively i was thinking that GUI can help with visualizing large lists and dependency graphs etc | 12:25 |
| minute | i hadn't seen cursorless at all! | 12:25 |
| hramrach | hope it's the correct name | 12:26 |
| minute | visualizing -> searching, filtering, drilling down, comparing | 12:26 |
| minute | a good TUI, though not a commandline tool because... TUI, for slicing and dicing data is visidata, but it is far from discoverable, it needs a lot of training to use | 12:27 |
| hramrach | why is it not discoverable? it's the same whether it's GTK or cureses, there eas even a GTK port to curses once | 12:28 |
| minute | hramrach: because you do things via keyboard shortcuts, there are no on-screen controls that you could discover | 12:29 |
| minute | keyboard shortcuts or a vim-like command line | 12:29 |
| minute | so it's as discoverable as vim | 12:29 |
| minute | or the aerc email client | 12:29 |
| hramrach | another nice piece of art was AutoCad. Not sure how it looks today but some vintage versions had customizable menu, and whenever you selected something in the menu it would type it in the commandline, and you could type it yourself as well | 12:30 |
| minute | yeah | 12:30 |
| hramrach | minute: however, there not being visible shotcuts is not property of cureses, it's property of that specific application | 12:30 |
| hramrach | eg Norton Commander and clones were very discoverable | 12:31 |
| minute | hramrach: i know that. my point was that visidata is a well working tui for dealing with complex data, but i would not say that an apt tui should be like this, because you'd have to memorize arcane shortcuts first to use it | 12:31 |
| minute | i also remember yast from suse | 12:32 |
| hramrach | oh, I haven't seen visidata, missed that point | 12:32 |
| minute | https://yast.opensuse.org/ | 12:32 |
| hramrach | I kind of use yast quite a bit | 12:33 |
| minute | (i used this like at the end of the 90s) | 12:33 |
| hramrach | aptitude as a TUI is kind of half way there - some features are easy to navigate, some are horribly crypic | 12:33 |
| hramrach | I use yast in the way 'I have absolutely no idea about this part of the OS, and there is a yast module for it so let me see if I can configure something with it before mucking with the low level details' | 12:35 |
| hramrach | but also there is a neat feature for copying partition table from one disk to another that would be otherwise 'easily' accomplished by an arcane sfdisk invocation | 12:36 |
| hramrach | it's something I even implemented in another tool but cba to write it | 12:38 |
| hramrach | My impressio is that people who learn excell or visidata to the extreme could just as well use python or whatever directly, perhaps with a library that has some conveninece functions for data analysis | 12:43 |
| hramrach | and wouldn't then produce epidemilogical studies with missing data because of excel row limit | 12:44 |
| hramrach | I think there is also a push to make 'professional' tools cryptic to sell trainings. | 12:50 |
| - mtm (QUIT: Ping timeout: 245 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 13:04 | |
| josch | if it were like that, i would not be surprised, seeing how crazy the vendor lock-in is. My local uni decided to upgrade thousands of PCs, throwing the old ones out, just because MS decided to make windows 11 non-functional on older chips... | 13:14 |
| * jn__ -> jn | 13:52 | |
| minute | reminds me of the SAP model. | 14:10 |
| hramrach | There is the thing that they might have done that because workarounds for features missing on older chips are annoying, doubly so when it's difficult to test chips with those missing features. Then again they may be in cahoots with chip manufacturers that want to sell more chips, difficult to say. | 14:18 |
| hramrach | The amount of code that goes into chip errata in Linux is non-trivial, and quite often the bug workaorunds also grow bugs of their own | 14:20 |
| + mjw (~mjw@gnu.wildebeest.org) | 14:22 | |
| minute | as a big institution who uses linux (or any FOSS OS) you can hire companies to fix it and keep it running for you on any hardware. with ms and apple stuff that is not possible. | 14:29 |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 15:10 | |
| - cobra (QUIT: Ping timeout: 264 seconds) (~cobra@user/Cobra) | 15:20 | |
| josch | "big institution" could even be a country as muncipalities in country X will all have the same processes and tasks to accomplish but instead of putting some money in local companies developing something that can be used by everybody, each muncipality is doing their own thing without sharing what they come up with... | 15:21 |
| josch | getting back on topic: | 15:21 |
| josch | can this be merged? https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/29 | 15:21 |
| josch | i tested the output of this on ls1028a and it didn't seem to break anything | 15:22 |
| minute | josch: yes please | 15:35 |
| + Guest97 (~Guest97@5.203.132.175) | 15:35 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@64.189.201.150) | 15:38 | |
| - Guest97 (QUIT: Ping timeout: 250 seconds) (~Guest97@5.203.132.175) | 15:49 | |
| + Guest97 (~Guest97@5.203.132.175) | 15:57 | |
| Guest97 | do we have an eta on the pocket reform being available through the webshop? | 15:57 |
| minute | Guest97: i think it is at least 3-4 months away | 15:58 |
| Guest97 | do we have an estimate of the retail price so i can start saving? | 16:03 |
| vkoskiv | What's the recommended way for switching between the system image kernel and one I compiled? I assume apt won't like it if I just `sudo make install` a kernel under /boot | 16:11 |
| - Guest97 (QUIT: Quit: Client closed) (~Guest97@5.203.132.175) | 16:11 | |
| + Guest97 (~Guest97@5.203.132.175) | 16:16 | |
| minute | Guest97: not yet, i'm sorry. component prices are volatile | 16:18 |
| minute | for example, display and ssd prices have risen significantly in the last months | 16:19 |
| - Guest97 (QUIT: Quit: Client closed) (~Guest97@5.203.132.175) | 16:23 | |
| vkoskiv | wifi issue seems to have something to do with sdio, it only happens on some boots, I get an error about sdio init in dmesg. I'll grab that error the next time I see it | 16:35 |
| josch | vkoskiv: when i test multiple kernels I cp it to a vanilla reform-system-image sd-card, adjust /boot/extlinux/extlinux.conf to pick up my uImage and boot my reform from that | 16:37 |
| vkoskiv | Okay, well, ethernet works now :D | 16:37 |
| vkoskiv | I just ran an apt upgrade now and it works. | 16:38 |
| vkoskiv | wifi too | 16:38 |
| vkoskiv | Though that may still have a random component to it, need to do more testing | 16:38 |
| josch | vkoskiv: what got upgraded? | 16:39 |
| vkoskiv | Let me grab that | 16:39 |
| vkoskiv | josch: https://mister-muffin.de/p/T0rj.txt | 16:40 |
| vkoskiv | Mostly userland stuff. Though now I had the ethernet plugged in while booting, let me try without. | 16:41 |
| vkoskiv | No, that's not it. Still works. | 16:42 |
| vkoskiv | Doubt it was the upgrade that did it, those are just userland libraries, mostly. | 16:43 |
| hramrach | tehre is also some make target for making packages in the kernel tree | 16:54 |
| josch | yes, bindeb-pkg | 17:21 |
| josch | vkoskiv: that is not a bad idea | 17:21 |
| josch | https://kernel-team.pages.debian.net/kernel-handbook/ch-bugs.html#s9.2.1 | 17:21 |
| josch | vkoskiv: you got a new advocate :) | 17:27 |
| vkoskiv | ? | 17:53 |
| vkoskiv | the reform lpc kernel module isn't loaded, maybe that's why I don't see battery state. | 17:54 |
| josch | vkoskiv: sorry wrong tab XD | 18:04 |
| josch | vkoskiv: yes, battery state in the DE depends on that module | 18:04 |
| + Christoph_ (~Christoph@p4fe7344e.dip0.t-ipconnect.de) | 19:45 | |
| + cobra (~cobra@user/Cobra) | 20:34 | |
| josch | minute: in case you missed it on the forums, https://mnt.re/system-image can link to a different branch than "main" | 21:31 |
| josch | for example right now it links to a job from the reform-system-imx8mp branch (as you can see from there being six system images) | 21:32 |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 23:01 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!