IRC Logs for #crux-devel Friday, 2018-03-30

Romsteri prefer stable and openjdk is working fine for my needs so i've since stopped using jdk00:02
j_vGlad to hear it. My main goal was to have a viable option that is opensource and worked at least as well. It's hard to gauge that it works as well, specifically since I can't afford the license for the testing harness to prove it out, but I haven't seen any breakage due to using openjdk instead of jdk when running stuff like minecraft or eclipse.00:06
Romsteri did try to build openjdk/ice tea ages ago but i gve up, i had no idea how that all worked. finding instructions on building it was near impossible.00:09
Romsteryeah been running minecraft with it just fine00:09
j_vI'm not going to brag about how long it took me to get the openjdk builds worked out to be where they are now... took a lot of work, a lot of looking at how others did it, and a lot of trial and error (i wish it had been less error)00:11
*** jaeger has quit IRC00:11
*** _________mavric6 has quit IRC00:11
*** xor29ah has quit IRC00:11
*** randn01z has quit IRC00:11
*** jue has quit IRC00:11
*** rmull has quit IRC00:11
*** nomius has quit IRC00:11
j_vdamn, wonder which side of the split i'm on00:12
Romstermy side :D00:12
*** nomius has joined #crux-devel00:13
*** _________mavric6 has joined #crux-devel00:13
*** xor29ah has joined #crux-devel00:13
*** randn01z has joined #crux-devel00:13
*** jue has joined #crux-devel00:13
*** rmull has joined #crux-devel00:13
Romsterand they are back00:13
*** maraku has quit IRC00:13
j_vwow, that was quick00:14
*** maraku has joined #crux-devel00:16
j_vok, pushed openjdk8 to contrib00:17
Romsterfixing texlive but i reported in 13 October 2017 and yet teK fixes other things and misses that since then. even when i attached a patch for it. -_-00:18
Romstersweet j_v00:18
Romsteri would love to see openjdk8 in opt someday and drop the closed source jre/jdk00:18
Romsterheck that was your bug report j_v for texlive.00:19
j_vYeah, teK closed my manpage fs report, even though I don't understand how a manpage being installed to /usr/share/man8/ could possibly be right... I really don't want to start an argument over it, but I'm still confused about his reply of 'works for me'00:20
Romsterhonestly i used of being yelled at for touching other peoples ports, and yet here i am fixing shit others leave broken.00:20
Romsterah what port was that on j_v00:20
Romsteri did see that but forget what it applied to00:21
j_va handful, just a sec, saved the email, just have to look at it00:21
Romsteri'll just fix em00:21
j_vFS#154500:21
j_vwas lsof, mplayer, rdiff, and xpdf00:23
j_vAll simple fixes00:23
Romsteryeah i'll go over them00:23
j_vI have an lsof fork, upstream just released 4.91, but I don't think it's CVE type update, though I'll check to make sure00:26
Romsterugh i'll be back in a sec, coffee and texlive compressing ina  single thread in xz is so damn slow.00:27
Romsteri used to use pigz for gz00:27
Romsterin pkgmk00:27
j_vnever tried it, maybe i should00:27
Romsterwhy can't libarchive have parallel compression00:27
j_vgood question00:28
Romsterthere was one for xz so it can use SMP but i can't remember the name00:28
j_vI know I've seen stuff on various mt compress/decomress progs, but can't remember names just now00:29
j_vbe back in a bit, gotta take of some home stuff for a bit.00:31
Romsterstill compressing...00:44
*** Workster has quit IRC01:03
j_vI'll have to delay pushing openjdk7... need to do some work tracking down why it crashes minecraft01:05
Romsterok01:07
Romsterwell i am using openjdk8 for minecraft01:08
j_vminecraft might not support 7, will look into it01:08
Romstercruxbot died01:23
Romsteri pushed the man page fixes01:23
*** jaeger has joined #crux-devel01:28
Romsterwell it works on openjdk8 so i wouldn't test openjdk7 with minecraft so much as other things01:31
Romsteri wasn't aware the testing framework wasn't free, that's such a catch01:32
Romsterah we still had a netsplit, cruxbot is back01:32
*** _________mavric6 has quit IRC02:42
*** _________mavric6 has joined #crux-devel02:43
*** xor29ah has quit IRC03:26
nomiusRomster: yeah, I saw the new deps, but even summing those are 1.3MBs...03:32
nomiusAnd the build was 10MB smaller...03:32
j_vAh, ok, minecraft >= 1.12 requires java8, so I don't need to go tracking that down. Still looking into eclipse, but the recent version I have runs on java 8 but not 703:35
j_vRomster, thanks for taking care of all those.04:16
Romsterno problem j_v04:26
j_vHey, I think that openjdk7 is ok to push, I'm just not convinced there's much call for it. Only kind of issue I foresee is complaints about it not working for things that require >= java804:31
*** xor29ah has joined #crux-devel04:31
j_vAlso, I'm planning to get started on openjdk9 tomorrow, though no guarantees how long it will take me to get a working port of it.04:32
j_vI'm going to play around using stuff like maven and gradle (and possibly jenkins) with openjdk7 tomorrow, will push openjdk7 if those work out.04:34
j_vBeing able to setup maven and gradle with openjdk would make setting up w3c Bootstrap a lot less painless.04:36
Romstereh i would keep openjdk7 in your j_v repo and just keep openjdk8 and later on openjdk9 in contrib04:37
Romsterj_v, https://gist.github.com/therealromster/d72f17088cb6fedf7c38fc23eef003e104:45
Romsterok you did add that hmm05:07
RomsterPkgfile: line 42: [: =~: binary operator expected05:10
j_vdamn, needs [[05:16
Romsterif [ -e '/usr/bin/ccache' ]; then05:20
Romsterthought i only gave you that bit05:21
*** chinarulezzz has joined #crux-devel05:23
Romsterif awk does not find ccache in the PATH it wont remove anything anyways05:23
j_vRight except it will add --enable-ccache to the options05:24
Romsteryep05:24
j_vhmmm05:25
Romsterunless you want the extra PATH check as a don't use ccache if it's installed but not on the PATH05:25
Romsterit could have CC and CXX set to ccache gcc and ccache g++ but i found that broke to many ports.05:25
Romsteror didn't work with some05:26
j_vI thought made sense when I did it, but now not so sure05:26
Romsteri just removed the second check to get it to build on my machine05:27
j_vi pushed the change from [ =~ ] to [[ =~ ]], but I can eliminate the second check if you think that's better05:30
Romsterif it works i am fine with it.05:30
Romsteri know not everyone uses ccache and it's a nich market.05:31
Romsterccache -> distcc over 8 boxens05:31
j_vI try forcing it either way to see if I can break it before changing yet05:31
Romsterif i am gonna find a bug in parallel make jobs it'll be me.05:32
j_v:)05:32
Romsteri like the work you have done05:32
j_vthanks. helps having good examples to look at05:34
Romsterbtw "mkdir -p $PKG..." can break when you are not using the usual umask 022, I've been defautling to "install -d $PKG..." as that by default is 0755 for a directory.05:35
Romsterelse if someone uses umask 0220 or something it'll make the footprint be out.05:35
Romsterwhich i am not but i have seen it crop up from time to time with other users.05:35
j_vgood point, I've been using the -m opt with install more, will review my ports to get that in there. also mkdir has the -m option also05:37
Romstertrue, but my opinion is because you can use install to install files as well it's uniform to use install -d05:40
j_vyeah, that's a good way of looking at. i can wrap my head around that.05:40
Romsterand i don't see it much but if you got a list of files to install i use install -t $PKG/foo file1 file2 file3....05:40
Romsterthat's just my opinion take it with a grain of salt05:41
j_vi don't mind salt05:41
Romster:D05:41
Romsterwhich ever way things are done keep it consistent i always say05:41
Romsterugh i need to stat on crux 3.4 building05:42
Romsterlate to the party as always05:42
j_vheh, it's a good method. also makes changing things easier later on.05:42
Romstertrue05:42
j_vi've got some 3.4 containers and a 3.4 vm... need to get an actual 3.4 install to disk05:43
Romsteri need to setup a docker 3.4, but i so forgot how to import a iso to a local docker container.05:44
Romsteri can use the test1 3.4 iso05:44
j_vi've never used docker... all my containers use my pkgwrap scripts and nsjail for the container manager05:45
Romsteri should start to use lxc sometime05:46
Romsteror was it lxd...05:46
Romstercrud, work and being to busy and i get out of touch with linuxy stuff05:46
j_vlxc/lxd i found really complicated, too much for my uses.05:47
j_vi wish that install had the equivalent to 'cp -a'05:48
Romsterare your scripts up anywhere?05:48
Romsteryeah sadly have to fall back to cp -a when you need that.05:48
Romsterinstall needs that option but you can do install -t destination *05:48
Romsterbut it wont create directories.05:48
j_vhttps://erebus.in/projects/pkgwrap for clone, https://erebus.in/cgit/cgit.cgi/pkgwrap for webview05:49
j_vI need to spend a little time with the HOWTO, which I've tried to keep up-to-date, but might be lagging a bit with all the changes I've done in last month or two05:51
Romstermore there than i was expecting05:51
j_vIt's been an wip for some time now, but been working pretty smooth for me05:52
j_vI usually only break it when I'm trying to do big changes, but also try not to push those kind of changes until I've tested them05:53
Romstermakes sense05:53
Romsteror you can have a unstable branch for testing stuff05:54
j_vThere's some downsides to my method... for one, it has to install packages into the container each time, which docker get's around by being able to do overlays via suid stuff05:55
j_vthere's supposed to be changes in the future to eventually allow non-privileged namespace overlay mounts, but when I don't know05:56
j_vThen I will work on setting up ability to have overlays for commonly large dependency builds, like things that depend on mesa3d and the like05:57
Romstertar -C /mnt/cdrom/rootfs.tar.xz -c . | docker import - romster/crux-3.405:57
RomsterError response from daemon: devmapper: Thin Pool has 95264 free data blocks which is less than minimum required 163840 free data blocks. Create more free space in thin pool or use dm.min_free_space option to change behavior05:57
Romsterugh well then i need to fix my space05:57
j_vah, yuck05:58
j_vchromium seems to build a bit faster with llvm/clang, though i'm not clocking it so that impression has no numerical backing05:59
Romster94.19% full -_-05:59
j_vyikes05:59
Romsterdocker rmi some old containers06:03
j_vWhat has me stymied on the ccache thing is how to determine when to add the --enable-ccache. I mean, me, I don't like idea of ccache, but I know it's just a personal preference; I just get concerned that stale object files could end up being used that should have been rebuilt from scratch (I know, paranoia).06:09
Romsterthat can't happen as the command line is md4 summed and so is the object. so if the command line does not match it does not match the built object and has to build and store the new object.06:11
j_vBut then it's no big deal to me, 'cause I'm not likely to ever have it installed. Now, distcc, that one makes more sense to me, because that just seems like extreme multi-threading06:11
Romsteri also do06:12
Romsterexport OS_CCACHE_COMPILERCHECK="$(/usr/bin/gcc -dumpversion; crux)"06:12
Romsterexport CCACHE_COMPILERCHECK="string:$OS_CCACHE_COMPILERCHECK"06:12
Romsterso it has to also match the gcc version and crux version06:12
j_vWell, I must admit, you definitely know more about it, and hearing what you are saying about it allays some of my distrust of the concept.06:13
Romsterthat's made it impossible to get a collision06:13
j_vvery cool06:13
Romsterexport CCACHE_PREFIX="distcc"06:13
Romsterso i feed ccache to distcc06:13
j_vnice, I really need to spend some time playing with that06:14
Romsteryou are more likely to break things with distcc if you don't keep the tool chain at the same version on all the nodes.06:14
j_vthat makes sense06:14
j_vOk, I think I'm seeing that that second check I added in openjdk8 was probably unnecessary.06:17
Romsterright if this works i can start on crux 3.4 testing06:19
Romsteri know i have gcc-fortran to bump06:20
Romsterthis isn't gonna be fun i can't use distcc on crux 3.4 docker container yet, until i update all my other nodes06:20
Romsteri was thinking of running a second distcc daemon on another port with a chroot for the new toolchain but effort.06:21
j_vWith 3.4, I'm going to have to revamp my gcj-jdk port since gcc7 has eliminated gcj06:22
Romsteroh06:22
Romsterso you can't compile java to native code anymore or another tool takes over that?06:23
j_vWill just bloat the gcj port, though. No help for it06:23
j_vI'll just use the gcc6 release branch as long as I can.06:23
j_vWell, I like the idea of bootstrapping all the way from a built java compiler, and openjdk needs java to build, so unless I can find another java compiler that is c or c++ based, I think I'm stuck06:25
j_vNot a big deal. Personally, gcj-jdk is mainly a bootstrapper for the openjdk ports. If anyone uses it for anything else that would be cool. But java 5 is pretty dated.06:26
Romstereh just keep building from the old version up.06:27
Romsteryour method to have a previous bootstrap is nice and saves most from having to build from so far back06:27
j_vright... it never sat well with me, forces everyone to have to bootstrap the whole shebang.06:28
j_vso current method i like better06:28
j_vs/forces/forcing/06:29
Romsteri like the idea of having ports to build the bootstrap still, i hate seeing a bootstrap binary and not knowing how it was created.06:29
j_vYeah, same here... one thing I wanted was to keep the transparency06:30
Romsteruser can chose what they want to do.06:31
j_vPeople can look at my ports tree and see how the bootstrap tarballs are created06:31
Romsteryep06:31
j_vNow, I just need to come with a openjdk-lite port for those who only want the minimum and don't want all the dependencies06:32
Romsteropenjdk8#8u161-b12-1.pkg.tar.xz' succeeded06:32
Romstersweet06:32
Romsterlike a headless jdk?06:32
Romsterthat be cool for a minecraft server06:32
j_vyeah, joacim mentioned an interest in something like that, I just haven't come with a method to do it yet06:33
j_vonly idea i've come up with so far is to start with the built package and strip it of unnecessary stuff and only depend on a few things06:34
j_vif crux had separate build-time and run-time dependencies, i could set up a separate build for it, but i'm not pushing for that (even though i'd be interested in seeing it)06:35
Romsteri had that idea but the more i keep thinking about it it would be to much effort for crux and arch does that already06:37
j_vyeah, so probably best bet is to find a solution that fits are own paradigm06:38
j_vI'll get around to creating a headless openjdk port, and do it by downloading a current pre-built openjdk package and work from there06:40
Romsteri don't think you can build a headless without X ?06:41
Romsterother than to build the whole thing as is then strip what is not needed, as you said.06:42
j_vas far as building it, yeah, i'm fairly certain without x is next to impossible to try to build (i've tried)06:43
j_vbut much it will run without it06:43
j_vs/much it/much of it/06:43
j_vthe mind goes faster than the fingers :D06:44
Romsterhehr06:49
Romstermmm i need a break07:08
*** onodera has joined #crux-devel08:02
*** john_cephalopoda has joined #crux-devel08:25
frinnstupdating crux.nu - i'll restart it soon'ish11:50
frinnstjaeger: seems rc.d/dropbear got overwritten, can you paste it somewhere from your backups?12:33
frinnstI assume it contains the options -s -p 2222 -w ?12:35
frinnstanyways, i've set it to that now12:38
frinnstchinarulezzz: can you send your public gpg and ssh keys?12:54
jaegerfrinnst: OPTS="-p 2222 -s" and $SSD --start --pidfile $PID --exec $PROG -- $OPTS13:47
jaegerare the only two differences13:47
chinarulezzzfrinnst: sure. where to send?14:00
frinnstmailinglist ok?14:22
frinnstor me, fredrik@rinnestam.se14:23
chinarulezzzfrinnst: sent to your email :)14:29
*** pedja has joined #crux-devel14:31
frinnstwelcome to contrib :)14:41
frinnstim gonna poke charlie again to get a vpn setup for ipmi kvm. I want to be able to upgrade crux.nu to 3.4 without having to drive an hour to do it on site14:50
chinarulezzzfrinnst: thank you! ^_^14:52
frinnstnp. just ask if something is unclear and you need help14:53
*** g0relike has joined #crux-devel17:18
*** g0relike-2 has quit IRC17:21
jaegerok, the duplicates view is much faster now, doesn't generate the list with every page load17:29
j_vjaeger: wow, that really speeds up portdbc's dup listing. nice.20:49
jaegeryeah, MUCH better now :)21:00
j_vI've been using your portdb and my  portdbc changes since we last talked about it. Working good.21:39
jaegergood deal21:52
jaegerProbably will add git repo support next time I mess with it21:53
j_vThat would be cool. Looking forward to it.22:02
jaegershouldn't be difficult... just like rsync and httpup, once it's synced locally it looks the same as the rest22:03
jaegergit is only *slightly* more involved due to checking if you're on the right branch or whatever22:08
*** onodera has quit IRC22:25
*** john_cephalopoda has quit IRC22:29
*** chinarulezzz has quit IRC22:57
*** xor29ah has quit IRC23:36

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