|
My $16 travel router / MIPS dev box is up and running. I added a $13 64GB microSD card for more storage. It has 128MB of RAM, which is enough to run real software like compilers. Power draw is 1.5W with the wifi off. code:
|
# ? Apr 8, 2024 21:21 |
|
|
# ? Jun 10, 2024 02:03 |
|
for all of those in here that had a bone to pick with dnf, it seems like dnf 5 will go into fedora 41 FESCo Approves The Fedora 41 Switch To DNF5
|
# ? Apr 9, 2024 08:01 |
|
Tankakern posted:for all of those in here that had a bone to pick with dnf, it seems like dnf 5 will go into fedora 41 I’m the nerd slap fight in the comments over the distinction between included and installed by default
|
# ? Apr 9, 2024 08:06 |
|
Tankakern posted:for all of those in here that had a bone to pick with dnf, it seems like dnf 5 will go into fedora 41 rewriting a package manager in c++ in 2024 seems such a weird move. as far as i understand a vm optimization thing to avoid a python install, because it for sure can't be *there* they have performance issues.
|
# ? Apr 9, 2024 08:11 |
|
why not? package managers do a lot of dependency resolution work
|
# ? Apr 9, 2024 08:31 |
|
can you really call that "work"
|
# ? Apr 9, 2024 09:02 |
|
Tankakern posted:why not? package managers do a lot of dependency resolution work but it'd be seriously idiotic to try to speed that up by rewriting the whole thing in c++, it is a single np-complete problem nesting in the middle of it, constant speedups is not really what you need. for one i'd trust a high-level code-base more to do a good ad-hoc smt solver, but really, if you're worried about it you should just pull in z3, because unless you're spending tens of millions on the rewrite you're not going to beat z3.
|
# ? Apr 9, 2024 09:33 |
|
ryanrs posted:My $16 travel router / MIPS dev box is up and running. I added a $13 64GB microSD card for more storage. It has 128MB of RAM, which is enough to run real software like compilers. do you feel you have achieved wireless freedom
|
# ? Apr 9, 2024 09:36 |
|
Cybernetic Vermin posted:but it'd be seriously idiotic to try to speed that up by rewriting the whole thing in c++, it is a single np-complete problem nesting in the middle of it, constant speedups is not really what you need.
|
# ? Apr 9, 2024 10:21 |
|
doing a check they're in fact not quite that dumb and both old and new dnf depend on libsolv for sat solving
|
# ? Apr 9, 2024 10:35 |
|
Tankakern posted:for all of those in here that had a bone to pick with dnf, it seems like dnf 5 will go into fedora 41 huh, i wonder if clear will switch too. it's not based on fedora, and swupd is the officially sanctioned package manager, but it's not very granular at all. intel has their own rpm repos set up, so at the very least i doubt that compatibility is going away
|
# ? Apr 9, 2024 17:11 |
|
also even though i sometimes run it on relatively garbage hardware i've never had any complaints about the speed of dnf (swupd is faster, but again you don't get as much control over what gets installed)
|
# ? Apr 9, 2024 17:21 |
|
dnf is just a package manager built on bad advice. do not gently caress? who does it think it is???
|
# ? Apr 9, 2024 17:22 |
|
Beeftweeter posted:debian is good because apt/dokg are good and more importantly pretty easy to use for new users fun fact: the .deb file-format has a size limit of about 10GB for any package, which comes from the choice of ar as the top-level container. Maintainers show zero interest in fixing this. 10GB was probably unthinkably huge in 1995 but in TYOOL 2024 it's really not all that big. I ran into it a few times when trying to package some commercial toolchains for internal company use.
|
# ? Apr 9, 2024 18:23 |
|
Poopernickel posted:fun fact: the .deb file-format has a size limit of about 10GB for any package, which comes from the choice of ar as the top-level container. Maintainers show zero interest in fixing this. huh, i didn't know that. them using extra compression on top of ar (which, lets face it, is not a compression format) doesn't change anything?
|
# ? Apr 9, 2024 18:35 |
|
Beeftweeter posted:huh, i didn't know that. them using extra compression on top of ar (which, lets face it, is not a compression format) doesn't change anything? That 10GB number is the the maximum compressed size of the data.tar.gz/xz member. It's a limitation of ar, which has a 10-character string field for the base-10 size of an archive member.
|
# ? Apr 9, 2024 18:51 |
|
Poopernickel posted:That 10GB number is the the maximum compressed size of the data.tar.gz/xz member. It's a limitation of ar, which has a 10-character string field for the base-10 size of an archive member. hmm interesting. i suppose that could be worked around by making separate debs for whatever huge thing you need installed (e.g. for your toolchain i guess you could break it up into the compiler itself, then i guess emulator? data, etc.)
|
# ? Apr 9, 2024 19:05 |
|
Poopernickel posted:That 10GB number is the the maximum compressed size of the data.tar.gz/xz member. It's a limitation of ar, which has a 10-character string field for the base-10 size of an archive member. New format: "V2" + 64-bit int = 10 bytes. gently caress bcd
|
# ? Apr 9, 2024 19:22 |
|
Beeftweeter posted:hmm interesting. i suppose that could be worked around by making separate debs for whatever huge thing you need installed (e.g. for your toolchain i guess you could break it up into the compiler itself, then i guess emulator? data, etc.) Hypothetically yes. I decided it was too hard, and not worth the benefit. I was trying to package an FPGA toolchain that has a 20GB install surface. Mailing-list advice was to make a top-level .deb with dependencies on smaller .debs. I would have needed to just grab random files until a .deb was filled, and then do that a few more times. Not very easy to automate, and also kind of yucky from a design perspective.
|
# ? Apr 9, 2024 19:30 |
|
ryanrs posted:New format: "V2" + 64-bit int = 10 bytes. yes, that would probably work. It would also break ar compatibility. Some people recommended options like that. Other people recommended having a data-1.tar.xz, data-2.tar.xz, etc kind of split instead of having a single data.tar.xz ar member. That would be more backwards-compatible. Ultimately the maintainers said "fuckit yolo, #WONTFIX" Poopernickel fucked around with this message at 19:44 on Apr 9, 2024 |
# ? Apr 9, 2024 19:33 |
|
eh i would've just packaged it in a compression format that can support an archive that large (...probably would've been xz, not now though lol) and then had it extract to the proper directories and add whatever config files were needed with a script or something
|
# ? Apr 9, 2024 19:34 |
|
Beeftweeter posted:eh i would've just packaged it in a compression format that can support an archive that large (...probably would've been xz, not now though lol) and then had it extract to the proper directories and add whatever config files were needed with a script or something debian: the best way to use our package-manager is "don't" Poopernickel fucked around with this message at 19:44 on Apr 9, 2024 |
# ? Apr 9, 2024 19:39 |
|
Poopernickel posted:debian: the best way to use our package-manager is "don't" well yeah, if you need a larger file size than it supports then you don't really have a choice
|
# ? Apr 9, 2024 22:09 |
|
Does it at least have a decent error message, not just "Error: ar returned 7, exiting."
|
# ? Apr 9, 2024 22:53 |
|
ar about to get 15
|
# ? Apr 10, 2024 00:08 |
|
Is there a way to completely erase any trace of a program having been on a linux system without having to track down every file and dependency that program ever created? I'm still banging my head against the wall with this drat torrent program (deluge) and as far as i can tell it doesn't work because a different torrent program (qbittorrent) I was using is causing deluge to malfunction because of different permissions qbittorrent set up.
|
# ? Apr 12, 2024 01:26 |
|
did you install qbittorrent with a package manager? if so, you should be able to just remove it e: but if you don't want to, usually a package manager will have some way of displaying all of the files a package contains along with the paths those files get installed to. you could then find out which one is the conflicting one and change its permissions Beeftweeter fucked around with this message at 01:35 on Apr 12, 2024 |
# ? Apr 12, 2024 01:31 |
|
Arson Daily posted:Is there a way to completely erase any trace of a program having been on a linux system without having to track down every file and dependency that program ever created? I'm still banging my head against the wall with this drat torrent program (deluge) and as far as i can tell it doesn't work because a different torrent program (qbittorrent) I was using is causing deluge to malfunction because of different permissions qbittorrent set up. well, “Linux system” isn’t really precise enough. Deb based you can do apt-get purge <package> to both torrent programs to remove the package AND any config then apt-get autoremove to remove any dangling dependencies. after that reinstall the package you want for rpm based distros lol and lmso afaik there is no equivalent to purge, even up to dnf. you can try rpm -e <package> with should remove config files at least the moral of the story here is sheets be using Debian
|
# ? Apr 12, 2024 02:01 |
|
Arson Daily posted:Is there a way to completely erase any trace of a program having been on a linux system without having to track down every file and dependency that program ever created? I'm still banging my head against the wall with this drat torrent program (deluge) and as far as i can tell it doesn't work because a different torrent program (qbittorrent) I was using is causing deluge to malfunction because of different permissions qbittorrent set up. yah it’s called containers
|
# ? Apr 12, 2024 02:18 |
|
akadajet posted:yah it’s called containers
|
# ? Apr 12, 2024 02:25 |
|
akadajet posted:yah it’s called containers i am quickly learning that lol. thanks for the help for my dumbass edit: that did the trick shitface! thanks a tonnnnnnnnnnnnnnne Arson Daily fucked around with this message at 03:00 on Apr 12, 2024 |
# ? Apr 12, 2024 02:53 |
|
Poopernickel posted:if you're a laptophaver, macos is still the way to go. Windows is next-best. Put your Linux in a VM (WSL is fine). Don't waste your life loving around with graphics drivers, wi-fi drivers, "oh, btw your camera won't work", bad battery life, etc. all I want for Christmas is to use Linux on a laptop with decent battery life Someone talk me out of trying asahi
|
# ? Apr 12, 2024 21:09 |
|
mila kunis posted:Someone talk me out of trying asahi no, asahi is fascinating and should be celebrated. the devs are insane. give them money.
|
# ? Apr 12, 2024 21:49 |
|
mila kunis posted:all I want for Christmas is to use Linux on a laptop with decent battery life both of the chromebooks i use get at least 8 hours, one gets 8-10 (has an oled and a faster processor) and the other (slower cpu, lower res lcd) gets around 14. if that's not decent battery life idk what is, 24 hours? e: they are not running chromeos. clear linux and the faster one dual boots windows 10, which gets about the same range Beeftweeter fucked around with this message at 21:56 on Apr 12, 2024 |
# ? Apr 12, 2024 21:53 |
|
mila kunis posted:all I want for Christmas is to use Linux on a laptop with decent battery life it's good op, probably the best linux battery life available, and it's not even its final form once thunderbolt support is done i want to convert all servers to macs for that sweet 10w idle on a max spec studio
|
# ? Apr 12, 2024 21:57 |
|
I haven't put Linux on a laptop in years because of the battery life thing. I wonder how much its improved since the last time I tried it. Linux desktop and server main tho.
|
# ? Apr 12, 2024 22:51 |
|
Beeftweeter posted:both of the chromebooks i use get at least 8 hours, one gets 8-10 (has an oled and a faster processor) and the other (slower cpu, lower res lcd) gets around 14. if that's not decent battery life idk what is, 24 hours? can I run docket and vscode and other crap on chromebooks
|
# ? Apr 12, 2024 23:34 |
|
vscode is just a web browser and embedded web page, it most certainly runs on a chromebook. if not natively, then the many many web-based implementations of it will work. dunno about docker but "real linux" on chromebook was some big news a few years ago iirc
|
# ? Apr 12, 2024 23:36 |
|
mila kunis posted:can I run docket and vscode and other crap on chromebooks i assume you mean docker, but sure? if you mean docket, idk what that is
|
# ? Apr 12, 2024 23:39 |
|
|
# ? Jun 10, 2024 02:03 |
|
definitely consult the coreboot device support list before buying one though
|
# ? Apr 12, 2024 23:41 |