IRC Logs for #crux-devel Monday, 2012-08-06

*** mike_k has joined #crux-devel01:28
*** acrux has quit IRC08:55
*** acrux has joined #crux-devel09:09
*** c0x` has joined #crux-devel10:40
*** c0x has quit IRC10:44
*** horrorStruck has quit IRC11:01
juejaeger: sorry :)11:11
jaegerno worries, I blame the downtime11:12
juewrt gcc, 4.7.1 has been released at 2012-06-14 so, I guess, we don't have to expect 4.7.2 before end of september11:15
jaegerSo it might be worth patching the ABI stuff in but they don't apply cleanly to 4.7.1, only a snapshot, I guess11:16
jueI'm not a big fan of it, but using a snapshot seems to be our best option than11:17
juein the hope that maybe the final 4.7.2 comes a bit sooner11:18
*** horrorStruck has joined #crux-devel11:19
jueif we are able to do a first rc end of this week, a final 2.8 might be ready end of september ...11:20
jaegerhopefully 4.7.2 will be out by then11:20
juewell, I'd say we wait for it with the final release11:21
jaegerI'm ok with that11:21
jueor even use 4.611:22
* jue hides11:22
jaegerI tested 4.6.3 quite a bit and it works fine but I'd prefer 4.7.2 if it's available11:22
jueyeah, me too ;)11:22
jaeger;a=shortlog;h=refs/heads/gcc-4_7-branch <-- should we just take a snapshot from the latest commit to the 4.7 branch?11:24
jueI thought about using 20120804 from
jaegerok, that's fine11:25
jueit's the currently the latest and a nice looking number :D11:25
juejust building it11:27
jaegeralright, I'll build that on my toolchain VM as well11:27
juejaeger: by chance I saw that ppl 1.0 is available, we are using 0.11.211:39
jueany reason for that?11:40
jaegerbecause it's what the gcc docs said to use11:40
jaegerthat may have changed with 4.711:40
jaeger doesn't list ppl anymore, now it's cloog and isl11:41
jaegerwill have to test that11:41
jaegerok, installed the right versions of isl and cloog, want me to upload those ports to a temp dir?11:51
jueyeah, please11:52
horrorStruckppl is still needed for graphite i think, are you planning to enable it ooc?11:53
jaegerI don't believe it is11:53
jaegerWe shall see11:53
pitillointeresting change11:55
jaegerwell, configure fails, bitching about PPL, so maybe they're both needed11:55
jaegerthat would be annoying11:55
juejaeger: thanks11:57
horrorStrucki had to patch gcc as it's looking for older verisons of ppl and cloog, not sure about latest snapshot (even if i built it yesterday :P )11:59
horrorStruckwell the two patches apply fine on latest snapshot so i guess they're still needed12:01
jaegerI wish I could find a clear document that says "you need this, this, and this"12:10
jaegergraphite seems like a mess12:10
jaegerI'm starting to think maybe it should be avoided12:10
horrorStruckppl, cllog and isl and you're done12:11
jaegertheoretically the ppl dep is supposed to be gone but obviously that isn't the case12:11
horrorStruckworks for me 9tm)12:11
jueit's a mess and principally I don't like such a lot of deps for gcc12:13
juebut, well :)12:13
jaegerwell, here's a question, then: do we NEED graphite? Is it gaining us anything tangible for the effort?12:13
jueI don't think that it's really worth the trouble, but many people asked for it ;)12:16
jueafter installing the old version of ppl and cloog to a temporary lib location gcc is compiling again, but I've to disable the version check for ppl and cloog12:21
jue(instead of patching)12:21
jaegermessy. =/12:22
juewell, installed gcc was broken because of the new sonames of ppl and cloog12:23
jaegerwhat the hell is the "choke me" bit supposed to be in the configure script?12:32
jaegerconftest.c:16:5: error: 'choke' undeclared (first use in this function)12:32
horrorStruckjaeger: just curious, did you chose the kernel that will be in 2.8 or it will just be the latest available when 2.8 goes live?12:37
jaegerIt usually ends up being the one that's current when the release happens, though in some cases I used an older one that was more stable or didn't have build issues12:38
horrorStruckok thanks :) last question and i'll stop being annoying, what about kernel headers for glibc build (really just ooc) ?12:39
jaegerwhat does ooc mean in this context?12:39
horrorStruckout of curiosity12:40
jaegernot sure which jue used in his port, I used whatever's in that toolchain dir... looks like 3.012:40
horrorStruckok thank you12:40
jaegerjue: My inclination regarding graphite is to leave it out for now. If the gcc devs aren't taking it seriously, it's to be avoided12:44
juejaeger: that's fine for me12:58
juejaeger: kernel headers I'M using currently ->
jaegerfor an rc are we just using your ports updates over the 2.7 ports?13:02
jueI'd say that we branch core/opt to 2.8 as soon we are ready with our toolchain13:04
jaegerI built gcc without cloog/ppl/isl for now, building glibc now13:05
jaegerThough I'll need to do it again for x8613:05
jaegeroh, where's the tzdata port? I'm missing that13:05
juejaeger: just one moment, I'll push my current changes13:09
horrorStruckjue: do you use any patch in your glibc ?13:12
horrorStruck(sorry guys i dont want to bother you, it's just that i'm interested, i find it very interesting to follow a new release development so closely and being able to speak with the devs)13:15
jaegerI don't mind the questions, just don't always have answers :P13:17
horrorStruckok thanks jaeger, sometimes i'm not sure if i'm being annoying or not :)13:18
jaegerjue: grabbed it from the git link you pasted in here recently13:19
juehorrorStruck: yes, one, as requested by frinnst13:24
jaeger$ prt-get listinst -v | egrep 'gcc|glibc|binutils|tzdata'13:25
jaegerbinutils 2.22-113:25
jaegergcc 4.7.2-113:25
jaegerglibc 2.16.0-113:25
jaegerglibc-32 2.16.0-113:25
jaegertzdata 2012e-113:25
horrorStruckjue: ok thanks13:26
jaegerjue: building on x86 now but I don't expect any trouble14:07
juejaeger: you've seen the patch for glibc?14:18
jaegerI hadn't until now, but I see your link above14:19
jueit's in the port in my repo14:19
jaegerat the moment just building gcc14:20
jueanother thing I'm not sure about ist the --enable-kernel=2.6.39, I think it doesn't hurt to set that to 2.6.32 or even .27 which are the oldest but still supported kernels14:27
jaegerI think there was a reason for 2.6.39... perhaps something udev required?14:28
jueyeah, that might be, will check14:28
jaegerwish I could remember14:29
juefrom udev's README: Version 2.6.34 of the Linux kernel ...14:31
juethe whole thing isn't critical, the only issue I see is that you cannot chroot into a system with a kernel older than --enable-kernel14:33
juebuilding of gcc just finished14:35
jaegerI'm content with leaving the required kernel version at 2.6.39, that would seem safe14:36
juejaeger: FYI, 2.6.39 is the kernel we ship with 2.7.1, 2.7 comes with 2.6.3514:42
jueso, if you have an old install with plain 2.7 you cannot chroot to it from 2.814:43
jaegeractually wouldn't it be the other way around? you couldn't chroot into the 2.8 from 2.7?14:47
juehmm, not sure :)14:50
jaegershould be simple to test, I guess14:51
jaeger2.7's glibc uses --enable-kernel=2.6.2714:51
jaegerso to me that sounds like any kernel 2.6.27 or later could be used14:51
jaegerso if 2.8 uses 3.4.7, for example, I would think that you could chroot into a 2.7 install14:52
jaegeron the other hand, if 2.8's glibc uses --enable-kernel=2.6.39, then a 2.7 install with a stock 2.6.35 kernel would NOT be able to chroot to it14:52
*** mike_k has quit IRC14:53
frinnstanybody got a grub 2.00 port?14:58
jaegerI think I do14:59
frinnst1.99, close enough :)15:00
jaegerIf I recall correctly the version number can be bumped to 2.0015:01
jaegerI just haven't uploaded it yet15:01
frinnstasdf 'gets'15:01
jaegersed -i '/gets is a/d' grub-core/gnulib/
frinnstmeh, failed on something else. unreadable output with -j8 :)15:05
juejaeger: didn't find anything useful about it, but I think you are right, to chroot into a 2.8 you need at least a system with 2.6.39 kernel15:07
jaegerfrinnst: hrmm, fails with flex problems now for me, no idea if I had run into that before... maybe that's why I never uploaded it15:10
jaegerjue: makes more sense that way, to me15:10
jueyeah, indeed15:12
juebtw, the wget update includes a 'gets' fix15:18
frinnstNEW       -rwxr-xr-x      root/root       usr/sbin/grub-sparc64-setup <- uh, 'ok'15:23
jaegerjue: the $version on line 42 of your gcc update should probably be 4.7.215:42
frinnstjue: I pushed a gtk3 port to opt, if you want to update transmission-gtk15:47
juejaeger: thanks, locally fix already, now pushed together with the footprint fix16:03
juefrinnst: great, will give a try tomorrow16:03
jaegerjue: it seems to be working fine for me here16:14
*** horrorStruck has quit IRC16:15
*** pidsley has joined #crux-devel19:00
*** joe9 has joined #crux-devel19:30
*** joe9 has quit IRC20:19
*** horrorStruck has joined #crux-devel21:05
*** mavrick61 has quit IRC21:30
*** mavrick61 has joined #crux-devel21:31
*** pidsley has quit IRC22:24
acruxinstead this is what i did on armhf, it's about a cruxppc toolchain evo22:50
acruxit's sad to see how many developers didn't update their libtool sources22:51
acruxand i'm still obliged to use very often  --build=$MACHTYPE22:51

Generated by 2.11.0 by Marius Gedminas - find it at!