IRC Logs for #crux-devel Friday, 2015-06-26

*** Feksclaus has quit IRC01:16
*** _______mavrick61 has quit IRC02:28
*** _______mavrick6 has joined #crux-devel02:30
*** Workster has quit IRC06:56
*** Workster has joined #crux-devel07:24
*** Workster has quit IRC07:50
*** Feksclaus has joined #crux-devel08:21
frinnstfuck, personal certificates seems broken in firefox :(08:22
frinnstcan anybody confirm?08:22
frinnstunable to login to startssl08:23
*** Feksclaus has quit IRC08:27
teK___aren't we forced to put the licensing crap into each and every Pkgfile?09:26
teK___arent we forced to ask each and every committer to {core,opt,contrib,xorg} ever for permission?09:26
teK___after all, you cannot put license restrictions on stuff you don't even (fully) own in the first place09:27
jueteK___: I'd say no at all, IMO it's enough to have the declaration in the handbook10:03
teK___that sounds doable..10:04
jueif that helps we can put one GPL license into the root of /usr/ports/{core,opt,xorg}10:05
teK___is there consent wrt the license itself?10:05
juewell, our handbook says so -> https://crux.nu/Main/Handbook3-1#ntoc510:06
teK___eek10:08
teK___GPLv3?10:09
teK___I'd strongly prefer public domain or bsd, but that's probably too late :)10:10
juethat's since ever in the handbook, guess that the link was pointing to GPLv2 in the past ;)10:11
Romsterfriggen licences... why can't it all be do wtf ever but you must provide the source of your changes.10:11
teK___any way.. the remaining issue is with contrib as there is no licensing statement for these ports10:11
Romstercontrib isn't even supported by crux right?10:12
teK___Romster: you see? That's why there is need for a license. I would not care for a second if a company would take my Pkgfile and make $$$ with it..10:12
Romsternothing much in crux that any other distro has already10:13
teK___that may be a bit harsh ;-)10:13
Romsterhehe10:13
teK___we have the $mojo of bloat-freeness10:13
Romsterwell there are a few others like void linux10:13
teK___Pkgfiles and their contents are a part of this :>10:13
Romsternot tried it10:13
teK___have you heard about that Arch-thingy?10:14
Romsterwhat arch thingie?10:14
teK___http://archlinux.org10:15
Romsterwhy call that a thingy....10:16
teK___:>10:16
teK___https://thisistwitchy.files.wordpress.com/2013/03/irony.jpg10:17
Romsteralso it seems i have the missing .footprint issues10:17
Romsterof course i know about arch linux... be more descriptive.10:17
Romsterthought you meant some /feature/ of arch10:18
*** nomius has quit IRC10:18
*** nomius has joined #crux-devel10:18
teK___and left it vague for the sake of irony, mate :)10:18
teK___did you bump harfbuzz-32?10:18
Romsterabout too10:19
jueteK___: btw, there's a new release of gnupg10:21
Romstersepen gnutls is at version 3.4.0 and it's now at 3.4.210:23
Romsterdon't know if he even reads irc logs.10:24
juefor sure10:24
Romsteralso --without-idn http://lists.gnupg.org/pipermail/gnutls-devel/2015-May/007582.html10:25
jueteK___: thx10:25
teK___thx.10:27
teK___I hope that committing things will be back to normal quite soonish (<5 weeks)10:27
teK___including regular runs of ck4up10:27
Romsteri've been a bit slow on updates myself.10:30
teK___btw does anyone here happen to know how we handle software 'requiring' ncurses with wide character support?10:32
teK___I am checking if we have support for that..10:32
teK___ /usr/lib* says yes. Yet we do not do the linking trick for the header files (as we do for the shared libraries)10:33
Romsterjust add in --enable-widec10:34
Romsteris it a problem to not have that?10:35
Romsteroh wait we do have configure for that... seems it's built twice.10:35
teK___to the port inquestion? It does not support that. It expects <ncursesw/curses.h> to exist10:35
Romster./configure $OPTIONS --enable-widec10:36
teK___I wonder if pinentry is the only port with that 'problem'10:36
teK___it _is_ enabled10:36
teK___but the C file says:10:36
teK___#ifdef HAVE_NCURSESW10:36
teK___#include <ncursesw/curses.h>10:36
teK___etc10:36
teK___we don't have that. :)10:36
Romsteryeah looking at the port it's got a funky way of installing files.10:37
teK___hehe10:38
jueteK___: I've filled a bug some times ago, see https://crux.nu/bugs/index.php?do=details&task_id=66710:42
teK___didnt see that :]10:44
Romsterif it's just pinentry, why not patch pinentry?10:48
Romstershould we be encouraging this split header include location?10:48
teK___as we are a very small distro, I guess we are not encouraging many things :P10:50
teK___except the coolness of having the $mojo described earlier :))10:50
jueok, FYI, no other problems besides reiserfsprogs with our new 3.2 tool-chain, looks very smooth so far :)10:51
teK___a quick grep says, five ports use ncursesw/10:51
teK___jue: awesome :)10:51
teK___Romster: 3 of these 5 are pinentry*10:51
teK___which I patched already.10:52
Romsterand the other 210:52
teK___stfl and sc10:52
Romsternot something i use.10:53
teK___those needed intervention in build().10:53
teK___Dunno about other software doing extended checks, possibly working around this issue themselves10:53
Romsterwhy do we need this wide stuff10:54
Romsterand why is ncurses set to compile twice.10:54
Romsterfor what narrow and wide?10:54
Romster1 and 2 byte chars?10:54
Romsteror is this more todo with urf8/16/3210:55
Romsterutf*10:55
Romsterjue, can the gold linker be included in binutils but not set as default?10:57
Romsterand possible for gcc to have LTO support10:57
Romsteror is either or both outside the KISS principal of CRUX?10:57
jueas far I can see, LTO is enabled by default11:03
juethough, no idea if it works11:03
Romsteri have used it in the past.11:03
Romsterit required another port for it to be compiled against gcc iirc11:04
jueTBH, I don't care about such specials ;)11:04
Romsterwell LTO is exotic i guess but i have seen some packages look for the gold linker.11:05
jueRomster: if you provide a patch against 3.2 we can consider11:05
Romsteri'll look into it.11:05
Romsterit is only used when -flto is on the gcc command line11:06
Romsterah that's it it needed graphite11:07
Romsteri'll look into this again...11:11
Romsteralso frinnst jue i just spotted11:11
Romsterhttps://gcc.gnu.org/bugzilla/show_bug.cgi?id=6588211:11
Romsterhttps://gcc.gnu.org/bugzilla/show_bug.cgi?id=6664711:11
Romsterhttps://gcc.gnu.org/bugzilla/show_bug.cgi?id=6603511:11
Romsterjue, yeah LTO is already enabled. just need the gold linker in binutils.11:13
Romstercan have ld and gold and prefer ld11:13
Romstersome packages call gold directly11:13
Romsterwell that last regression is for fortran so it don't affect core/gcc11:15
jueRomster: it's useless to look for gcc bugs right now, because a final 3.2 will have at least gcc 5.1.111:17
Romsteroh right it probably is a bit early.11:18
jueyeah, reiserfsprogs fixed too, thanks Ubuntu for the hint :)11:19
Romsterfrinnst, wouldn't this be better https://projects.archlinux.org/svntogit/packages.git/tree/trunk/freetype26.patch?h=packages/firefox than adding ac_add_options --enable-tree-freetype to firefox?11:22
Romsteri also see firefox 38.0.6 but i guess that's not important11:23
frinnstRomster: its not needed for firefox 39 or whatever the version will be so I wont bother11:44
frinnstit should be released any day now11:44
Romsterah ok11:55
frinnst38.0.6 has not been announced yet either11:55
Romsterstrange11:55
frinnstnot for mozilla :-)11:56
Romsterheh11:56
teK___nvoke the pass as the last stage before code generation (at the same time the existing cookie-based stack protector pass is invoked).12:13
teK___AsylAsylhätten0 Kommentare Video12:19
teK___asdf, sorry :)12:20
jueok, finally updated glibc-32, hope that works, cannot test it because I don't use any 32-bit stuff13:34
juefrinnst: I totally forgot about gcc 4.8.5, any problems with it?13:35
frinnstnot yet13:38
frinnst:-)13:38
frinnstI also dont use any 32bit stuff at home so Im also unable to test13:38
jueok13:40
jueyou are following core/3.2?13:41
frinnstyes13:41
frinnstusing the git driver13:41
juegreat13:41
juesorry, I didn't increase release= with each commit, so you should track for chances manually13:43
frinnstyeah noticed that no worries. Dont think we should keep changing $release in that tree for now anyways13:44
jueok13:45
mechaniputerGuys, I've been looking at how other distros handle licensing of ports collections and build scripts. At this point I'm relatively sure that we just need to add a copy of whatever license into the root dir of the collection and we're fine. Gnu's GuixSD distro does this with their build scripts, for example. Might not hurt to aks the FSF also, but I'm not a dev and it isn't my decision.13:50
mechaniputerHope this helps.13:50
juemechaniputer: yes, it helps, thanks for looking into that13:57
jueRomster: done -> https://crux.nu/gitweb/?p=ports/core.git;a=commit;h=64af57b694706633d8aa583cc0e77c4d024a47a415:21
Romsterjue, nice15:30
mechaniputerThe only other catch that I have thought of is possibly needing a copyright at the top of each file, but I think that's more of an advisory than a requirement. To be safe, you could add a copyright file along with the license file.17:16
Romsteri don't see gentoo or arch doing that to every ebuld/pkgbuild file.17:33
mechaniputerNo, which is part of why I think it's an advisory more than a rule.17:35
*** Feksclaus has joined #crux-devel17:37
mechaniputerHowever, arch doesn't currently seem to have copyright of their PKGBUILDS since they seem to have ignored the licensing issue for theirs. It's ambiguous and there's a bug report on it from awhile back that remains unresolved. That's possibly a huge problem for parabola.17:38
mechaniputerI know gentoo to be in violation of the GPL since they don't have a copy of the GPL in their tree anywhere.17:39
mechaniputerBut, CRUX's situation is much more trivial to correct than Arch or Gentoo.17:41
jaegerprologic: do zfs/spl 0.6.4 build with 4.0.x and 3.18.x?17:42
frinnstprologic: do you plan to keep linux-firmware updated? its getting quite old'ish I think17:59
jueperl updated to 5.22.0, all tests passed ;)18:08
jaegernice18:09
juebesides the man-page location task, everything should be in place now18:09
jueI'd suggest to do that as late as possible, to avoid a lot of merge problems18:10
*** Feksclaus has quit IRC18:28
juejaeger: would be nice if you could take a look at the 32bit stuff, I don't use that and cannot test if everything is ok18:39
jaegerglibc-32 specifically or something else?18:41
jueyeah, glibc-32 and some tests whether we are able to build 32-bit versions with our tool-chain or not18:43
jaegera very simple test would be a hello world18:43
jaegergcc -m32 -o hello32 hello.c, something like that18:43
juewhatever you want, I mainly thought at the ports in compat-3218:55
jaegerI'm not running a 3.2 toolchain machine yet but I'll try to put some time into it this weekend19:03
jaegerI will be out of town monday through wednesday next week19:03
jaegerlooks like syslinux still doesn't do UEFI very well so I guess I'll leave the grub config on the ISO alone21:27
jaegeror maybe it needs more core modules21:30
prologicjaeger, I'll test23:30
prologicfrinnst, I'll update it thanks :)23:30

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!