libera/#devuan-dev/ Thursday, 2020-05-21

onefangHeisenbugs we call them.  When you try to observe them, they change.00:21
fsmithredWe should hire Schroedenger's cat to hunt the Heisenbugs. They're good at getting bugs.00:53
onefanglol01:11
golinuxSeems there is still intelligent life on this planet.  :D01:41
rrqfsmithred: yes, the squashfs comes pre-populated with /sys files01:50
rrqno forget that (it's early morning here )01:51
fsmithredyours or mine?01:51
fsmithredno, there's nothing in /sys in the squash or in the initrd01:51
fsmithredoh01:52
fsmithredI know that's true for refracta isos01:52
fsmithredchecking...01:52
fsmithredempty in desktop-live01:53
rrqyes I confused myself01:54
masonTomorrow's meeting pad: https://pad.dyne.org/code/#/2/code/edit/qfcxCPswV-LR4N2YOy0h8zgO/02:26
rrqfsmithred: how about a Donald Duck synchronization (aka sleep) just before initrd.img:scripts/init-bottom/udev:15 ?04:02
onefangBut then people will complain that sysvinit takes too long to boot up.  Waaaaaaah!04:05
rrqin theory it should be possible to only wait until all hotplug events have been dealt with, and maybe a "udevadm settle" would do that04:09
bgstack15Regarding gzip, xz, and bzip2, I kind of prefer gzip but mostly because that's the oldest and most common. Once or twice have I tried to pick a compression for one of the optimized features (minimal size, or minimal decompression time, or minimal compression time) but mostly gzip is good enough.13:47
bgstack15https://www.rootusers.com/gzip-vs-bzip2-vs-xz-performance-comparison/ is a great read.13:47
yetiwhat about safety?14:21
yetihttps://www.nongnu.org/lzip/xz_inadequate.html14:21
yetithis aspect mostly is overlooked14:22
yetiI've no idea about the correctness of the claims in there but speed and compression factor sure is not all to look at14:23
yetigzip just compresses too less for some stuff.  If I can shorten downloads thousands of times by using xz instead of gz, that multiplies up to lots of bandwidth less and lots of faster downloads (relative to the unpacked volume)14:25
yetiI had no fail with xz so far but if I ever understand lz being safer ore more futureproof or such, the less less than 1% less compression relative to xz I see on larger stuff like os install images sure is ok...14:27
yetioooh haaa... less less less... :-)  typing long stoff in a single scrolling line (weechat) considerered harmful!14:28
yetitime to go erc...14:29
yeti(somewhen... ommmmm...)14:29
FrapeXhi all15:33
FrapeXdoes devuan beowulf i386 support systems without the SSE2 instruction set? I scanned the binaries/libs of the live dvd (using analyse-x86.pl), and found out that there are plenty of binaries and libraries that contain SSE2 instructions.15:33
FrapeXThe kernel itself is compiled for GEODE LX and up.15:39
yetiSSE2 instructions you see in sleeping code might not be executed on all platforms15:40
yetilike   if(hasSSE2) SSE2CRC(data) else portableCRC(data)15:41
yetithe only problem I remeber (years ago) was flash starting not to run on PIII any more (which I didnt care much about)15:43
FrapeXyeti: that makes sense. But for example the latest firefox 68ESR is loaded with SSE2 and CPU's without SSE2 arent supported anymore.15:43
FrapeXs/SSE2/SSE2 instructions/15:43
* ShorTie wonders, https://git.devuan.org/devuan-doc/documentation/tree/wip/new-packaging-guide/maintainers gives me 404 errors for guides at bottom15:44
FrapeXyeti: yea I remember flash, only Flash Player 11 worked on non SSE2 CPUs.15:46
FrapeXIs the Devuan i386 tested on real hardware without SSE2 support? I know from experience that even if you set QEMU to not support SSE2, it still executes them without error. So kinda hard to test?15:51

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!