IRC Logs for #crux Monday, 2018-06-18

*** abenz has joined #crux00:01
*** onodera has quit IRC00:48
*** tilman_ has joined #crux01:31
*** abenz has quit IRC02:08
*** _________mavric6 has quit IRC02:42
*** _________mavric6 has joined #crux02:43
*** maraku has joined #crux05:14
*** abenz has joined #crux05:16
*** smolboye has joined #crux05:37
*** abenz has quit IRC05:44
*** abenz has joined #crux06:11
*** abenz has quit IRC06:18
*** parlos has joined #crux06:20
*** smolboye has quit IRC06:20
*** maraku has quit IRC06:34
*** abenz has joined #crux06:44
*** jue has joined #crux07:06
*** jue has quit IRC07:06
*** jue has joined #crux07:06
*** abenz has quit IRC07:21
*** de11 has joined #crux07:22
*** abenz has joined #crux07:23
*** smolboye has joined #crux07:31
*** smolboye has quit IRC07:32
*** smolboye has joined #crux07:33
*** fun has quit IRC07:38
*** abenz has quit IRC07:53
*** smolboye has quit IRC08:02
*** smolboye has joined #crux08:02
*** smolboye has quit IRC08:05
*** smolboye has joined #crux08:05
Romsteryeah i built qt5 on a spare box i am now using instead of docker currently.08:20
Romsterso i didn't see the docker issue.08:21
*** jue has quit IRC08:28
*** smolboye has quit IRC08:53
*** smolboye has joined #crux08:55
*** de11 has quit IRC09:06
*** SiFuh_ has joined #crux09:52
pedjait's interesting to hear 'I don't trust AMD' from technical people10:05
Romsteri am more i don't trust intel10:05
Romsterbut i like intel ssd's10:06
ryuopedja: haha. so, Intel is *more* trustworthy?10:06
pedjaapparently so :)10:07
Romsterwhile i use my 7 year old amd system that's barely got any venerability's other than spectre10:07
ryuothat's nuts.10:07
Romsterbut that's only minor and fixed in software10:07
Romsterand not to the same degree as intel's spectre bugs10:07
pedjaI was surprised to hear that form people that really should know better10:08
ryuoIt's quite possible, though, that AMD only has fewer vulnerabilities... due to Intel patenting their more flawed approach.10:08
ryuoI read that when Meltdown was first revealed.10:08
Romsterand all eyes are on intel currently so who knows what other bugs of amd sparc risc arm has other vunabilites not disclosed yet10:09
Romsterbut fuck intel... what were you thinking to check the registers after the horse has bolted.... for permission to do something. where amd checks it before.10:11
*** abenz has joined #crux10:12
Romsterwhat logic enen maks that seem like a logical choice giving read access to say someones credit card then checking if you had permission to read that?10:12
ryuoThat's weird. I used advpng to optimize an icon set...10:12
pedjatrusting any corporation is silly10:12
ryuothe final file size is *larger* in bulk? What gives?10:12
Romstereven makes*10:12
pedjaryuo, you are probably right. because of Intel's patents, AMD had to get creative10:21
pedjaIntel was the top dog for far too long, and it shows.10:22
*** abenz has quit IRC10:22
*** smolboye has quit IRC10:22
ryuopedja: yea... i was experimenting with recompressing an icon-pack and i was shocked at what I found...10:23
ryuoin bulk, they took up more space on disk than before.10:23
ryuoI was like, how can this be?10:23
*** smolboye has joined #crux10:23
ryuoThey should be using fewer disk sectors.10:24
*** abenz has joined #crux10:24
pedjacompare imginfo output, maybe there is some clue there10:24
ryuopedja: have you used advpng before?10:24
pedjano. never heard of it before10:25
ryuoit's part of advancecomp.10:25
ryuoit incorporates some different algorithms for compression DEFLATE/ZLIB based formats.10:25
ryuoin the case of advpng, it can apply zopfli, the currently known best method for optimizing them.10:26
ryuoit's great for reducing file size for PNG files.10:26
ryuobeyond the -9 setting.10:26
ryuobut, not practical for all kinds.10:27
ryuomostly useful for largely read-only images.10:27
ryuoi was needing to package an icon set as part of a GTK theme, so i thought...10:27
ryuothis isn't size-reduced. should I try optimizing it?10:28
Romsteri remember the days i tried to lzma and other jar (not java but another compressor) lha arj and others files that were already highly compressed like mp3 and flac and jpg etc.... and not gained much if anything made the archive bigger than the uncompressed files.10:30
Romsterdue to the meta header and crc sums etc.10:30
ryuoRomster: yea... this makes the .tar.xz more compact, at least, because the PNGs are better compressed.10:30
*** onodera has joined #crux10:31
Romstermaybe the icons are already highly optimised?10:31
ryuoRomster: i doubt it. if they weren't, advpng would just ignore them.10:31
ryuoit doesn't overwrite them if the new file is larger or equal size.10:31
Romsterwhat about optipng?10:31
ryuoRomster: used both. same result.10:32
Romsterthen i guess the program that made them png's did a good job already10:32
ryuonot individually.10:32
ryuoto be honest though, i expected not much to change.10:33
ryuomostly because of how each file is small and has to take up a minimum amount of disk space.10:33
ryuoSo, recompression for *disk* space doesn't make sense.10:34
ryuoOnly for data transfers.10:34
Romsterjust a straight tar of then would make sense for transfer purposes.10:36
ryuoRomster: yea... it's just weird. the uncompressed tar of the old one is smaller, but the compressed tar of the new one is smaller.10:36
Romsteri would imagine the only time you could compress a entire directory of png's would be if the pall;ate of each png was exactly the same set of colours used in the png10:37
ryuomy theory was recompressing the files would reduce the final package size due to smaller PNGs.10:37
Romsterpalate code table10:37
ryuoRomster: ah. advpng does a better job than optipng in some cases.10:37
Romsterbut the chances of that being the same are very small10:37
ryuoparticularly when optipng can't reduce color depth.10:38
ryuozopfli is rather effective...10:38
ryuocan be used on .gz or .zip files too.10:38
*** abenz has quit IRC10:38
ryuonot really worthwhile there.10:38
Romsteri beleve it does some lossy changes like if you got 2 almost identical colours you can make that 1 palate entry and reduce the file size some be using the 1 colour than 2 that's almost identical and barely identifiable by the human eye?10:39
ryuoyou're usually better off with a better algorithm for archives.10:39
ryuoRomster: no, i don't think it does.10:39
Romsterlong range windows for compression like lrzip10:39
Romsterheader files are a good candidate for compression as they all have the same licence text at the start of a set of them.10:40
*** smolboye has quit IRC10:40
Romsterand in source code10:40
Romsteri don't know all the ins and outs of how pngs work and compressions10:41
*** smolboye has joined #crux10:41
ryuoRomster: you know how advpng works, yes?10:41
Romsterbut i do know it has a table of colours and it refers to that for each pixel.10:41
ryuoRomster: all it does is optimize the compression algorithm used on the image data.10:41
ryuoit uses better compressors that still yield a compatible stream.10:42
ryuozopfli is the best known to date.10:42
Romsterthere is a few different zlib implementations even10:42
ryuoSo... knowing that, it shouldn't make a set of PNGs larger on disk.10:43
ryuoor in aggregate.10:43
ryuoi wonder what's up.10:43
ryuoi wonder if this is ZFS being weird again.10:43
ryuoi should try this again on a regular FS...10:44
pedjaspeaking of compression algorithms
Romsternice pedja10:50
ryuoMystery solved.10:53
ryuoadvpng was recompressing *symbolic* links.10:53
ryuook, time to omit them.10:54
Romstersurely there is a argument to omit symlinks10:54
ryuoyea, to find.10:54
Romsteror did you just do foo * blindly :P10:54
ryuonah, i used find.10:55
Romsterto glob everything10:55
Romster-type f10:55
ryuonow i should see results.10:55
ryuobecause, logically, a recompressor should result in <= original file size.10:55
Romster- the overhead10:57
Romstercrc or FEX or what have you and the table of where the files are located if you do tarball it up.10:58
Romsterso can not type10:58
ryuoRomster: even with that overhead...10:59
ryuoRomster: incidently, i think this is why CPIO is used for initrds.10:59
ryuoand due to minimum file size sectors... recompressing icons doesn't yield much space saved on disk, though it can reduce the cost of transferring files.11:02
Romstertrue a 1KiB file still takes up a 4KiB FS block, but some FS are smart and can pack these small files in the same inodes11:03
Romsteror well directly into an inode until it becomes bigger to better fit a FS block11:04
ryuoyea, but i don't think ext4 is that advanced.11:04
ryuoi'm just surprised recompression isn't used on more icon packs.11:04
ryuoit's an easier way to reduce file size of already compressed files11:05
ryuointerestingly, debian/ubuntu still doesn't use compressed kernel modules.11:05
ryuoThough, support for it has been available for awhile now.11:05
pedjacool, next libarchive release should support zstd11:05
ryuoi introduced that into FW and we saw our on-disk footprint shrink a lot.11:06
ryuoIt reduces the disk space needed to store the kernel.11:07
ryuoBut i imagine CRUX has less use for that given how you promote customized kernels.11:08
Romsterkernel doesn't use that much space as it is, but it would be a benefit for embedded11:08
ryuoWhich reduces the usefulness of kernel module compression.11:08
ryuoRomster: it gave around 60-70 more MBs of disk space.11:08
ryuoa modest amount.11:08
Romsterwhat i would like to see is SMP support in libarchive why does it not support parallel compression/decompression11:08
ryuoRomster: it does, in a parallel dimension. :P11:09
Romsteri used to use a patched pkgmk with pigz instead of gzip and god damn it was fast when it hit big archives like texlive firefox qt stuff.11:10
Romsterlibreoffice too11:10
Romsteryou wait ages for the compression11:10
Romsteron a single darn core.11:10
Romsterin a SMP system11:10
ryuoRomster: filthy pigz! =p11:11
ryuobut anyway.11:11
Romsterpxz could be used for lzma but why not in libarchive.11:11
ryuoyou do sacrifice one thing by using parallel compression.11:12
ryuolarger files.11:12
Romstera tiny bit of compression11:12
Romsterbig whoop11:12
Romsterfor packages use.11:12
ryuosome would say, if you care about raw speed, why not use lz4?11:12
Romsterbecause pkgadd does not read that? or maybe it can with libarchive hrmm11:13
ryuoyea... no idea.11:13
Romsteri don't fee like getting dirty in code after work11:13
ryuo.tar.xz is still better if you're planning to store.11:14
ryuoor transfer.11:14
ryuoi would prefer a faster option for test builds.11:14
ryuowhy waste time on expensive compression methods for packages you're not going to keep?11:15
Romsteri'd rather sacrifice like 5% compression for 4 times the speed for a 4 core system11:15
ryuodepends on what you're doing i guess.11:16
ryuoi tend to want to maximize compression.11:16
ryuoat least when the data in question isn't volatile.11:16
Romster-15 use 2petabytes for the window :P11:16
Romsternot in our lifetime11:17
ryuoi find it strange why early BASIC used line #s...11:18
ryuoalways seemed like an impractical way to program because it makes it harder to revise code later.11:18
Romster10 print "hello world!"11:19
Romster20 goto 1011:19
Romsterand then walk away11:19
Romsterthe idea was to leave enough space to go oh i can add a 5 something line and then i got 1-4 and 6-9 left in case i need more lines between.11:20
Romsteri had to in the past even renumber my entire program because i forgot something. that was not fun.11:20
Romsteri think it was because text editors were not a thing back then11:21
Romsterand basic needed a way to know what line you were on11:21
Romsteri am not a programmer though so....11:22
pedjaRomster, for bsdtar to support lz4, liblz4 should be added as a libarchive dependency11:24
pedjaadding lz4 as an option to pkgmk then shouldn't be that hard after that11:25
Romstermore core dependencies i doubt it's worth the effort?11:25
Romstermaybe for a hack to a users own system but the benefit vs the cost.11:25
Romsterbad enough we need macko for mesa now. like wtf11:26
Romsteron the other hand the choice of using elfutils for revdep was a good one ryuo as elfutils is in core now. for the kernel11:27
Romsterso that worked out well11:28
ryuoI see.11:35
*** john_cephalopoda has joined #crux11:53
*** SiFuh_ has quit IRC11:58
*** SiFuh_ has joined #crux11:58
*** SiFuh_ has quit IRC12:24
*** SiFuh_ has joined #crux12:34
SiFuh_I have a btrfs problem :-/12:36
SiFuh_UUID does not match fsid12:36
frinnsthuh? so what happens?12:37
frinnstpaste paste paste12:37
SiFuh_No idea12:37
frinnstdidnt know the FS actually cares about the uuid12:37
SiFuh_neither did I12:37
frinnstwhat does btrfs fi sh say?12:40
SiFuh_hold on, still making pastes for ya12:44
pedjahow dangerous is to set UUID with btrfstune -U $UUID?13:08
SiFuh_pedja: I tried that13:09
pedjaI guess it didn't work?13:09
SiFuh_Nah  hang on I will run again and show you the result13:10
SiFuh_ERROR: fsid 47814679-712c-48e1-b470-5baebe95049c is not unique13:10
pedjaSiFuh_, perhaps ask in #btrfs?13:22
SiFuh_good idea13:24
*** nthwyatt has joined #crux13:25
pedjafrinnst suggested it when I had an issue with btrfs on openSUSE VM13:26
*** darfo has quit IRC13:27
SiFuh_I got a nice reply ;-)13:33
SiFuh_darkling> THat's pretty special. Never seen that before.13:33
*** SiFuh has quit IRC14:42
*** smolboye has quit IRC14:44
*** nthwyatt has quit IRC14:44
*** smolboye has joined #crux14:46
*** darfo has joined #crux14:47
frinnstheh yeah no kidding14:56
frinnstvery odd14:56
frinnstthe #btrfs people are really nice to deal with14:58
frinnstdarkling is a support beast14:58
john_cephalopodafrinnst: There are some channels that have quite some knowing and helpful people who answer fast.15:00
john_cephalopoda#nouveau, the channel for the open-source nvidia drivers, has several people who are really helpful.15:01
frinnstand then you have channels like #crux with unhelpful angry people like me in them :>15:03
john_cephalopodaWell, #crux is pretty quiet most of the time. It takes some time until anybody answers or says something.15:05
*** Workster has quit IRC15:05
*** ranvgufaxccukcpw has joined #crux15:05
*** Workster has joined #crux15:06
pedjaand some people asking the question go away after 30 secs :)15:25
john_cephalopodaHello I got problem xy can anybody help?15:27
john_cephalopodaACTION quit15:27
pedjasomething like that, yes15:27
john_cephalopodaThere are bots that remember which users are known. When a new user joins, it says "Welcome <username>, please stay around a few minutes after your question, nobody might be around at the moment"15:28
Anselmosome people ask a question, wait for an hour or two and still get no response15:30
Anselmoand then leave :P15:30
SiFuh_john_cephalopoda: We know you problem has somethign to do with bots. ;-)15:31
*** smolboye has quit IRC15:31
*** smolboye has joined #crux15:31
pedjathese seem sensible
john_cephalopodaUsually people don't stick around long enough to read the topic text.15:33
SiFuh_well frinnst and pedja  we still have no solution to my btrfs problem, even though I have a full functional rc version now, with kdaves patches specific to my problem. ;-)15:34
pedjatime to bug your MEP's, EU citizens
pedjaor this
pedja(sorry for the spam)15:48
SiFuh_pedja: problem solved and a new update will be released by the btrfs team ;-)15:53
SiFuh_frinnst: read above too ;-)15:53
pedjanice. pretty impressive work by kdave15:54
SiFuh_yes very much so15:54
SiFuh_I didn't tell them I had a backup on LTO5 ;-)15:54
pedjaso, after all this, would you recommend btrfs as a filesystem :) ?15:55
SiFuh_pedja: I was actually thinking about how it reminded me of when reiserfs came out.15:56
SiFuh_They were very helpful and that alone was enough for me to change to their fs. I lost a lot of data in those days.15:57
pedjareiserfs was all the rage for some time, iirc15:57
SiFuh_pedja: I do like btrfs, that is why 2 of my backdrives run it.15:58
SiFuh_the rest run on ffs ufs ext4 and on tape15:58
pedjain hindsight, I should've chosen xfs instead of jfs, which is basically unmaintained at this point. oh, well15:59
SiFuh_i used xfs in 200315:59
SiFuh_always data issues15:59
pedjawhat is ffs?16:00
darfoFor Fsck's Sake?16:01
pedjaI know it in that context, had no idea you could store the data in it :)16:02
darfoThe Berkeley Fast File System16:02
pedjaah. Unix[tm] ?16:03
darfonot sure. it seems to be available on FreeBSD, NetBSD, and OpenBSD16:03
SiFuh_it is an OpenBSD filesystem16:04
joacimi dont remember if i used xfs or reiserfs when i had issues with files disappearing16:04
joacimxfs on linux doesnt have the more fancy features anyways16:04
pedjaI had that issue with ext416:04
joacimi think it has a lot more features on aix16:04
pedjafile system should be as boring as possible, imho16:05
SiFuh_filesystems are like a made bed pedja. I think beds should be as boring as possible too ;-)16:06
darfoI've had xfs on my pvr since 2007. Numerous power outages. Journal replay always seems to work but a very active mysql table got corrupted sometimes and needed repair.16:07
darfoHave rural power so outages are common. Finally bought a ups for it and no more problems (yet).16:08
darfoafter herding that pvr for so long I think they should be as boring as possible too16:09
pedjamodern file systems have some cool features (snapshots, compression), thou16:12
joacimxfs does that too i think16:20
joacimor maybe thats just the fancy volume manager some systems use16:20
pedjaits development picked up some speed, by RedHat mostly, iirc16:25
*** tsaop has joined #crux17:03
tsaopjaeger: containerd needs git and btrfs-progs as dependencies, it seems17:04
jaegertsaop: I'll take a look, thanks for the report17:07
*** parlos_ has joined #crux17:16
jaegerseems like both runc and containerd try to run git during the build but neither of them fail without it. containerd does indeed need btrfs-progs, though, I'll add that17:20
cruxbot[contrib.git/3.4]: containerd: added missing btrfs-progs dependency17:22
*** smolboye has quit IRC17:24
frinnstSiFuh_: problem with btrfs-progs?17:26
*** smolboye has joined #crux17:26
frinnstffs is an amiga fs! and it will always be!17:26
joacimamiga ffs!17:27
frinnst <317:28
*** parlos_ has quit IRC17:30
*** smolboye has quit IRC17:39
pedjawtf, Youtube?
pedjaBlender Foundation videos, too17:54
john_cephalopodapedja: What happened? My browser isn't liking youtube much, can't look at the link properly.17:55
pedjaYT blocked their videos for, as of yet unknown, reasons.17:56
pedjatheir p1racy bot is gone berserk17:56
*** brian|lfs has quit IRC17:56
*** de11 has joined #crux17:59
*** smolboye has joined #crux18:08
*** de11 has quit IRC18:27
joacimand this is why i youtube-dl stuff i think is important18:29
*** JanC has quit IRC18:35
*** JanC has joined #crux18:39
frinnstpeople are putting too many eggs in the youtube-basket19:08
frinnstits a google product - it might not even be here tomorrow :)19:08
dbrookeI recently upgraded my main desktop machine to 3.4 (which went pretty smoothly so thanks to all involved) and now have Firefox 60.0.1 which no longer seems to support a middle click paste of a URL from the X selection into the main pane to replace the current page in that tab. \19:15
dbrookeSearching has failed to reveal a way to restore that behaviour so does anyone know of a way?19:15
darfo"(UFS; also called the Berkeley Fast File System, the BSD Fast File System or FFS)" from
darfoI wonder which came first? BSD or Amiga?19:23
onoderaI always hated that functionality dbrooke :p19:23
onoderadbrooke: you could try out the following about:config setting19:24
dbrookeheh, it's how I always opened a link from email, IRC etc.19:24
onoderaI don't knmow if this does what I think it does though19:24
dbrookethanks, I'll investigate that19:25
darfoaction facepalms19:25
darfonever new firefox could do that before. would have saved a lot of fiddling with xclipboard19:25
dbrookeonodera: that fixes it, thanks19:27
dbrookeI was looking for paste in about:config and didn't find it ...19:27
onoderaI was lucky and looked for middlemouse :p19:27
darfomiddlemouse.paste seems to paste the current selection19:28
dbrookenow just to sort out manual geolocation and I'll be happy (well, less unhappy 8-)19:28
*** SiFuh_ has quit IRC19:32
*** SiFuh has joined #crux20:04
*** dlcusa has quit IRC21:00
*** tsaop has quit IRC21:02
*** pedja has quit IRC21:08
jaegerI love this keycap set, but holy shit, that's an expensive keyboard -
jaegermainly the solarized one21:13
joacimlooks nice21:23
joacimthe coloured legends makes a lot of difference21:23
joacimEstimated ship date is Dec 5, 2018 PT.21:24
joacimtakes too long for their stuff to ship21:24
joacimby the time it ends up here, i'd have forgotten about it already21:25
*** onodera has quit IRC21:32
*** dlcusa_ has joined #crux21:34
*** dlcusa_ has quit IRC21:38
*** dlcusa has joined #crux21:39
*** Workster has quit IRC21:48
*** Workster has joined #crux21:49
*** abenz has joined #crux22:08
*** smolboye has quit IRC22:12
*** abenz has quit IRC23:07
joacimwould be nicei f they focused on making a single pack like that. mass produced to make it cheaper23:37
frinnsto bought a new sofa 2 months ago. forgotten about it until today when I was told it had arrived23:46
*** john_cephalopoda has quit IRC23:47

Generated by 2.14.0 by Marius Gedminas - find it at!