IRC Logs for #crux-devel Wednesday, 2019-06-05

*** deus_ex has joined #crux-devel06:55
ryuorofl. i had a funny idea to try in another channel with a lot of programming bots.07:47
ryuotrick them into activating each other with their output statements. causing a cycle between them.07:47
ryuobut i won't do it. spam.07:47
*** Workster has quit IRC08:34
*** wspwbxfefahntttv has joined #crux-devel08:34
*** xcusmmjarnriegmo has joined #crux-devel08:44
*** Workster has joined #crux-devel08:51
TimB_tek__: any chance you are free this week sometime?09:00
*** jue has joined #crux-devel09:57
*** stenur has joined #crux-devel10:47
juejaeger: is iso.git uptodate? I'd like to prepare our ChangeLog14:32
jaegerHrmm, probably not. I'll look at it today. Sorry about that, I always forget to maintain it while working on a new release14:33
juedo we have any feedback wrt rc1?14:35
jue.oO, we have a new shiny cruxbot :)14:39
jaegernone that I've received via email. Have seen a couple of comments about it in #crux that seemed positive14:45
juesounds good, how many downloads do we have until now?14:48
jaeger612 hits in my nginx access.log but probably some of those are automated or spiders, etc.14:50
juewow, much more than I thought ;)14:59
jaegeryeah... it would be interesting to know how many of those are valid user downloads15:11
jaegerjue: ok, updated iso.git15:20
jaegerNo problem15:21
jaegerregarding the new cruxbot, it's pretty basic, but since it only has one job, hopefully it will work fine :)15:21
pedjaissues I've seen reported with 3.5 on the ml are mostly with non-official repos. kde, that kind of thing.15:22
jaegerspeaking of which, I'm running into a lot of packages failing to build if vala is installed15:24
jaeger(while building MATE)15:25
jaegersome are MATE packages, some not, like libsoup, etc.15:25
pedjacontrib/vala is seriously out of date15:35
pedjaone of tek__'s, I think15:35
jaegeryeah. I'm going to test if updating it helps15:36
pedjaI had no issues using 0.42.2, fwiw15:36
jaegerdid you have to install libgvc for 0.42.2?15:36
jaeger0.45.1 is squawking about it15:36
pedjano. what, another dependency? yay15:37
*** dlcusa has left #crux-devel ("Leaving")15:38
*** dlcusa has joined #crux-devel15:40
pedja0.42 is EOL. time to update to 0.44, see what breaks15:41
jaeger0.44.4 seems to want it as well. graphviz support15:48
jaegerdoesn't look like it has an option to disable15:48
pedjabuilt fine here, which makes sense since graphviz is installed.15:51
jaegerI just installed graphviz, testing again :)15:52
jaegercan you build libsoup properly with vala 0.44 installed?15:52
pedjalet me try. a sec.15:52
jaegerNice, it works here. So updating vala might fix all the others as well, I'll have to test15:53
pedjalibsoup builds fine here, too15:54
jaegerok, good, thanks15:54
juebtw, I'm out of town beginning next Monday for around 10 days, shall we wait with the release until I'm back or do it this week?15:54
jaegerI don't have a strong preference. What's yours?15:55
juehmm, good question :)15:59
jueif we use rc1 it's probably better to do the release soon, so the packages on the ISO are not too outdated, but well ...16:01
jaegerI could easily build another during the time you're gone, if you prefer. I would also update the kernel, probably.16:03
juefrinnst: what's your opinion?16:06
*** stenur has quit IRC16:32
jaegergiven that there are a bunch of updates for packages on the ISO already since rc1 I'd kinda prefer to build a newer one for the release19:38
jaegercmake, curl, firefox-bin, openssl, vim, xorg-server amongst others19:39
pedjanew mesa should be out soon too19:41
dlcusaAs there is nothing new on FS#1667, is removing pkg-config, installing contrib/pkgconf, and adding "pkgconf: pkg-config" to /var/lib/pkg/prt-get.aliases the proper upgrade for those packages needing prtconf?19:47
ryuoWas that even implemented?19:59
pedjaaliases? a long time ago. rarely used, thou20:17
TimB_aliases work fine, and pkgconf works fine for me as well dlcusa20:19
pedjayou are both braver then me, I'll give you that :) I stay away from messing with core ports20:21
dlcusaMy alias "rust: rust-bin" is rather helpful, pedja.20:21
pedjadlcusa, thanks for reminding me to add that20:21
TimB_pedja: come over to the dark side20:22
TimB_I feel like aliases should be used more where it makes sense20:23
pedjaTimB_, I was referring to pkg-config->pkgconf :)20:23
TimB_openjdk is another example, j_v is having a look20:23
TimB_pedja: I know :D20:23
pedjaI use aliases all the time20:23
pedjaah. a bit slower then usual I am, then :)20:24
TimB_haha, no worries :)20:24
TimB_dlcusa: how is libreoffice working for you? All good I hope? :)20:24
dlcusapedja, I think that is going around right now--I'm sure fighting it, anyway.20:24
dlcusaIndeed, TimB_20:25
pedjaheh. I am kind of old, so20:25
TimB_great, happy to hear that :)20:25
dlcusaI'm half a year from 70, myself.20:28
pedjareally? cool20:32
dlcusaIn some ways, yes--in others, not so much.  ;-)20:32
frinnstI'm all for a quick release20:42
frinnstpkgconf is a bit "meh" for me20:45
frinnstiirc we looked at that a few months back or so20:45
TimB_I don't really see a difference to be honest - which also means it just works. if it really is faster.. it's a nice touch?20:47
pedjawhat are the numbers on 'faster' claim?20:52
pedjacouple %? 2,3x?20:53
pedjaalso, afaik, it's not bug-compatible with pkg-config :)20:54
ryuomy thoughts were it should be put on hold. if it's really that great, wouldn't most distributions eventually adopt it? i know this doesn't always work (systemd) but it seems like sound reasoning in this work.20:54
pedjaryuo, agreed20:54
ryuoif Debian adopts it, i suspect it's a sign it's a drop-in replacement with no real issues.20:55
TimB_I wouldn't disagree either20:55
ryuoit has it already packaged but afaik it's not the default.20:55
TimB_pedja: I had read something about speed somewhere, but I can't find it right now.. not sure, I'll post a link20:56
pedjaspeed is not that important in this case, hitting a bug some random app triggers because it depends on some weird pkg-config wart is :)20:58
ryuoso, do you folks think I should try to backport some of my optimization techniques i found for pkgutils?20:59
TimB_so far, I thought I had exactly that once, tried it, didn't resolve it. so I changed back and still not a single issue. I have rebuild everything multiple times with it :)20:59
ryuoi think i can backport my optimization to make footprint generation faster.20:59
pedjaryuo, and when you do, I'll test it with my cuda port :)21:00
TimB_sounds good21:00
ryuoeven if the library ends up being a bust, i'd still like to backport some changes that don't require a massive overhaul.21:01
ryuook, well, back to work then.21:02
pedjaisn't someone rewriting pkgutils in Go?21:06
pedjaI think I see that mentioned a few times21:06
ryuoi forget who, but that's not a viable project for official use i imagine,.21:06
ryuoi doubt they'd ever go for a Go compiler in core.21:06
ryuoplus Go isn't exactly an ideal choice for library purposes.21:07
ryuoI got the feeling it's optimized for use for applications and services.21:07
pedjaGoogle is behind it, so probably21:08
pedjaa bit weird build system, tbh21:09
pedja'portable' binaries are nice, thou21:09
pedjaif bzip2 gets rewritten in rust, as is the plan apparently, rust has to be in core then. ouch.21:11
TimB_rust package bins out at 133M here :)21:14
TimB_ouch indeed21:14
pedja168Mb rust-bin package21:14
pedjahm. developer that rust-ed librsvg maintains bzip2 now.21:15
TimB_the only thing I noticed is if gtk icons don't work anymore, just rebuilding librsvg won't do it anymore21:16
pedjayou have to re-run gdk-pixbuf-register, afaik21:23
pedjarebuilding gdk-pixbuf cache is already in the post-install for librsvg. hm.21:25
pedjaI guess some people don't run those, so have issues with svg icons.21:26
pedjasomeone on the ml did recently.21:27
*** stenur has joined #crux-devel21:31
ryuohm. smooth progress so far.21:31
ryuoif this works it should cut the time required in half.21:31
TimB_pedja: didn't know that, but I run scripts usually so this can't be it entirely :^ not sure what broke them in the first place to be honest21:33
pedjaTimB_, no idea either21:35
TimB_it's not that important anyway :) but rust really seems inevitable21:36
TimB_with bzip2 (interesting read btw), more will follow I guess21:36
stenurOn NetBSD they are talking about and seem to use svrg-c since some days, i see it fly by.21:57
stenur(Luckily i do not have anything like that here.  Just watching NetBSD.)21:57
pedjathat's interesting, only svg fork I saw other then librsvg was using rust too21:58
pedjanot that big of a deal with rust-bin. annoying, yes, but :)21:58
pedjaTimB_, that seems to be trend. switch to meson, switch to rust21:59
stenurThat is what i don't like. Problems are enough wherever you look.22:01
stenurChanging build system and complete rewrites in other languages .. well.22:02
pedjanot an issue for the majority of people using binary distros. they don't care/know :)22:02
prologichmm? what's going on?22:29
ryuoand finished it seems.22:38
ryuoprologic: well, i'm doing minor modifications to pkgutils to improve its cpu performance.22:39
ryuohm. with the mods i've done, we can also drop the sort pkgmk does.22:42
ryuoor at least in theory, since i sort the metadata i read.22:42
ryuoand yep. time needed cut in half.22:43
ryuonow to test the output results.22:43
prologicI have to ask ... who cares? : )22:45
prologicits a tool that you run to install a pacake22:45
prologicnot performance critical in any way :)22:45
prologicbut OTOH it is fun optimizing things :)22:45
ryuoand that additude is why we have people making crap that's super slow in JavaScript.22:45
prologicI'm with you on that one :)22:46
prologicBut that's just because JS is a frankly terrible language :)22:46
prologicin any case ++ nice one :)22:46
ryuowell, i was reviewing how pkgutils works so i could make it into a library and found some mods i could backport without serious changes.22:46
ryuoefficiency improvements.22:47
prologicthat's the way :)22:47
prologicinstead of rewriting it; learn and understand it and improve it :)22:47
ryuoor both even.22:47
ryuoACTION shrugs.22:47
prologicI still recall when it was all pure shell scripts :)22:47
ryuohuh. guess the shell script sort method is still important to maintain compatibility.22:50
ryuook. i won't be removing it as i can prove it's not identical.22:50
ryuotime to package this up as a nice git commit.22:51
prologicI really like pkgutils as a package manager :)22:52
ryuojue, frinnst, jaeger, TimB_, pedja: I have finished a patch to pkgutils to improve footprint mode's time performance. Please review and test, but I expect it to behave like the old algorithm did in terms of output.
prologicwhy aren't you sending a patch email to the devel list ?23:04
prologichard to inline review a paste on some random website :)23:04
ryuoprologic: laziness.23:21

Generated by 2.14.0 by Marius Gedminas - find it at!