IRC Logs for #crux-devel Tuesday, 2020-01-14

*** xor29ah has quit IRC00:39
*** xor29ah has joined #crux-devel00:44
*** erol has joined #crux-devel00:55
*** Workster has joined #crux-devel02:19
ryuojaeger: i had a weird idea to reduce the ISO image some but wouldn't reduce RAM usage.06:18
ryuojaeger: i was researching bootloaders. they can accept multiple initrds. one idea i had was to make a premade base rootfs tarball that could be used for both booting the iso and installing part of the system.06:19
ryuobut probably not viable.06:19
ryuobut would be for squashfs06:20
ryuoyou could then overlay any modifications over it.06:20
ryuoand just copy the contents of the squashfs for the base install.06:20
ryuothat could save space by removing redundant packages.06:20
*** pedja has joined #crux-devel08:27
frinnstone way to do it. not sure it would simplify stuff09:49
Romsternet inst and pull in packages?09:50
ryuofrinnst: yea. i've seen some ISOs that design their rootfs to be the foundation of the installation.09:50
Romsterthe other idea would be rootfs and a pre-populated base database of all the files in each package but stored as a single archive of base09:51
ryuoso, you're suggesting building the real rootfs on the fly?09:51
frinnstthe squashfs idea I think is to just not decompress the filesystem to a ramdisk09:52
ryuoindeed, but it could do a bit more.09:53
ryuobut then CRUX's iso isn't that huge.09:53
Romsterthink jaeger is concerned that that the amount of ram it uses to be about to eject the disc is getting up high09:54
ryuowas just thinking a design change might make it practical to bundle the base image as a squashfs image that can be copied from for installation as well.09:54
Romsterfor older or embedded stuff?09:54
ryuomostly just to save space.09:55
ryuothere's redundancy in packaging the base components also used in the live cd.09:55
Romsterthat's like ~100MiB peanuts09:55
ryuowell, still. it's a huge slice of the current image size.09:56
ryuoit would reduce it around 15% or so i think09:56
ryuoafaik there's still ways to load a squashfs into RAM.09:57
ryuoubuntu live CDs can do it if you specify a parameter for it09:57
ryuoi've used that when i need to disconnect the boot drive, usually due to a low # of ports.09:58
*** xor29ah has quit IRC11:11
*** xor29ah has joined #crux-devel11:13
*** pedja has quit IRC11:50
*** pedja has joined #crux-devel11:53
jaegerryuo: I've played with the rootfs foundation thing in the past. It's interesting but never spent a lot of time on it14:59
ryuojaeger: i see.14:59
jaegerRomster: it's only partly the RAM cost, because lots of RAM is common these days... but more that it annoys me14:59
jaegerI've also considered building the entire boot env out of busybox with just enough support to have networking and static pkgutils, etc.15:00
ryuojaeger: it's not a problem for x86, but i imagine it could make issues for ARM.15:03
ryuomost systems have 2G or more these days on x86.15:03
jaegercrux-arm doesn't use it15:03
jaegerWith that said, using more than 1GB RAM for an install rootfs seems kinda silly. We're not quite there yet but getting closer15:04
pedjaif you put rootfs on a diet, with just the basic tools/busybox, wouldn't that mean that it can no longer be used as a rescue live system? it15:17
ryuoquite possible pedja.15:17
pedja(sorry, I missed the discussion so far)15:17
ryuospecialized tools may not be present.15:17
ryuolike FS stuff.15:17
jaegerin its minimal, yes. But it would also be possible to add specific tools15:18
pedjawell, that would suck, imho. I've used the iso more times for that then for actual install15:18
jaegerI do that frequently as well, so don't worry that it's not being kept in mind15:21
*** Romster has quit IRC19:09
*** Romster has joined #crux-devel19:10
*** pedja has quit IRC21:07
*** prologic has quit IRC21:20
*** prologic has joined #crux-devel21:21
*** frinnst has quit IRC23:28

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!