=== asac_ is now known as asac | ||
keesj | Hi, I want to get started on ubutunu-arm. I currently want to target qemu-arm versatile (possibly using the arm11 cpu) and would like to use 9.10 and rootstock. Would this be recommented (elinux.org has a "ok" turorial when using beagle + rootstock the topic links on this irc channel point to what looks to the old way of doing things) | 08:26 |
---|---|---|
lool | keesj: this is incompatible | 10:04 |
lool | keesj: It can work with a patched versatile kernel though | 10:05 |
lool | keesj: I think there is such a kernel in the development release now, or you can get some binary ones in some places | 10:06 |
lool | keesj: Just FYI, the real hardware named "versatile" which qemu emulates only supports up to v5 ATM | 10:07 |
armin76 | lool: question: now that karmic is the latest released version, does that mean that jaunty doesn't get any update anymore? | 11:03 |
lool | armin76: In terms of where developers spend their time, I don't think many stable updates will happen in jaunty, it's not impossible but unlikely; but of course security support continues and if critical bug fixes need to be addressed in jaunty they can be | 13:26 |
lool | armin76: Even at this point, karmic is much less of a focus than lucid | 13:26 |
armin76 | lool: ack, thanks | 13:49 |
asac | ogra: yes ... my install feels good too ... thats t2 maybe?;) | 14:04 |
asac | fta: chromium codecs fails on i386? | 14:28 |
asac | wanted to copy it | 14:28 |
fta | asac, please don't. | 14:28 |
fta | asac, http://code.google.com/p/gyp/issues/detail?id=123 | 14:29 |
asac | sure | 14:29 |
asac | i wont copy if the build fails ;) | 14:29 |
asac | fta: would that impact arm? | 14:29 |
fta | asac, last comment | 14:29 |
fta | probably yes | 14:29 |
asac | why did you switch to make just now :( | 14:30 |
fta | i thought it was ready | 14:30 |
asac | can i disable that manually again? | 14:30 |
asac | e.g. just a single rules line? | 14:30 |
fta | I don't have common.gypi in this package so most variables have no default during the build (like target_arch, chromeos, branding, library, and guess what.. cflags and ldflags), so i set those in GYP_DEFINES, but it doesn't work for cflags/ldflags. The build fails with an asm error (the fix is to add the missing -O2) but if i set CFLAGS=-O2, it breaks all the defines and the -I. | 14:31 |
fta | nope, it will be the same with scons (unlike for chromium) | 14:33 |
fta | asac, ^^, testing my proposed fix right now | 14:41 |
asac | what changed that even scons suddenly doesnt work anymore? | 14:43 |
fta | scons would work if i also ship the whole src/build/ dir from chromium in the codecs package (and probably other dirs), while i just took the ffmpeg.gyp file, which is enough, in theory | 14:45 |
fta | i just loose the defaults, so i have to specify them as i said above | 14:46 |
fta | what happens if is say gcc -O2 -Os ? which one is used? | 14:47 |
fta | -is+I | 14:47 |
asac | feels not right to "loose defaults" ... they might do arch dependent tweaks etc. | 14:48 |
asac | i dont think we want to duplicate all defaults in rules | 14:48 |
asac | feels like not running configure | 14:48 |
fta | that's the whole point of this gyp, not run configure | 14:48 |
fta | gyp is a configure rival | 14:49 |
fta | in python :S | 14:49 |
fta | and upstream is dropping support for scons anyway, so the sooner i move, the better | 14:50 |
asac | right. but i expect that the defaults contain detection stuff etc. | 14:50 |
asac | yes. but you dropped the defaults ... which feels like a bug | 14:50 |
keesj | lool I will be able to create a kernel if needed. still rootstock + qemu-system-arm should be easy right? | 15:00 |
fta | asac, i'm pretty conservative here. missing defaults make gyp fail, so it FTBFS for us and we notice. And i set only those: GYP_DEFINES="target_arch=ia32 chromeos=0 branding=Chromium library=static_library use_system_yasm=1". The only problem is cflags/lflags that dpkg-buildpackage wants to overwrite | 15:08 |
fta | asac, if i really want to pull src/build too, that would mean 3 svn to pull to create the tarball. | 15:12 |
fta | asac, i already name my stuff 0.5+svn20091210r34297+34315, i can add a 3rd rev-id :P that would mean rebuilding more often (for any commit in src/build, even totally unrelated) | 15:13 |
fta | i should start a contest of the longest version ever used for a package | 15:15 |
lool | keesj: Yes | 15:35 |
keesj | thanks | 15:48 |
asac | fta: isnt it like "extra_cflags=..:" ? | 15:49 |
fta | asac, no, not without src/build :( | 15:51 |
fta | asac, chromium-codecs-ffmpeg_0.5+svn20091210r34297+34315+34433.orig.tar.gz uhuh | 16:21 |
asac | self.WriteLn("$(OBJS): CFLAGS := $(CFLAGS_$(BUILDTYPE)) " | 16:22 |
asac | shouldnt that be | 16:22 |
asac | self.WriteLn("$(OBJS): CFLAGS += $(CFLAGS_$(BUILDTYPE)) " | 16:22 |
asac | ? | 16:22 |
asac | in make.py | 16:22 |
asac | (gyp) | 16:22 |
fta | won't work, as make CFLAGS="-g -O2" would overwrite it anyway | 16:23 |
asac | += ? | 16:23 |
fta | yes | 16:23 |
fta | asac, http://paste.ubuntu.com/340609/ | 16:24 |
asac | yes thats ok too | 16:24 |
fta | asac, i'm done fixing the codecs | 17:47 |
fta | asac, you can copy the codecs now, they are all green in lucid. you need to copy gyp too (a new build-dep) | 19:10 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!