irclog2html for #oe on 20041016

00:01.38mikelpthat may have been the problem, the symlink is working now
00:02.33mikelpgrrr, no wget
00:09.49CIA-203CoreDump 07 * r1.2369 10packages/ (9 files in 9 dirs): Merge
00:09.49CIA-203CoreDump 07 * r1.2351.3.1 10packages/ (24 files in 16 dirs): Prepares the coming split of opie-18n packages into seperate files
00:11.10*** join/#oe zimbo (~bonis@pD9E38602.dip.t-dialin.net)
00:12.34mikelplol
00:12.55mikelpoe gets quite upset when the uname -m result has a space in it
00:24.49kergothheh, there, implemented rudamentary copy on write for the dictionary i'm playing with
00:25.29mikelpwhere do we set STAGING ?
00:25.47kergothlook in packages/conf/oe.conf
00:25.53kergothyou can override any of those in your local.conf
00:26.00kergoth(or any other .conf that oe.conf includes, for that matter)
00:26.24mikelpI'm going to have to override
00:26.31mikelpthe default is getting a space in the name
00:26.52mikelpwhich upsets install -file $STAGING_BINDIR a lot ;)
00:28.03mikelpto override STAGING needs a full path or just relative to TMP_DIR ?
00:28.25kergothoe expands all variables. use what you want
00:29.04mikelpIf nothing is set, I assume its calling uname ?
00:29.38kergothread oe.conf
00:29.44mikelpbah :p
00:29.47kergothspecifically the BUILD_OS, BUILD_ARCH variables
00:31.03mikelpI see no BUILD_OS ;)
00:31.10mikelpbut I'll go hunt it down
00:31.15kergothlook again
00:31.19kergothline 26
00:31.23kergothheh
00:31.33mikelphuh
00:31.38mikelpoe/conf/oe.conf ?
00:31.41kergothno.
00:31.47kergothpackages/conf/oe.conf
00:31.49mikelplol
00:31.55kergothoe finds things by looking in OEPATH.
00:32.05kergothif your OEPATH has packages first, which it almost certainly does, thats what its using
00:33.47mikelpaha, I need to override BUILD_ARCH
00:36.28mikelpActually "getting OE" working on PPC/OSX might not be a bad talk
00:38.30*** join/#oe magnumbi (~magnumbi@ip68-9-104-137.ri.ri.cox.net)
00:38.37kergothheh
00:39.10*** part/#oe magnumbi (~magnumbi@ip68-9-104-137.ri.ri.cox.net)
00:39.59kergothi just added a list of value/conditional pairs to class oevar.  the applyconditional() function walks the list, at each value, it takes the condition, which is actually a python statement, eval()'s it, and if it succeeded, replaces the default value with that.
00:40.04kergoth:)
00:40.18mikelpdeep joy. os x cp doesn't support -a
00:40.38kergothheh
00:40.41kergothoemake coreutils-native
00:41.43servo888Is this a bug? http://www.pastebin.com/110693
00:41.58servo888"| : undefined reference to `main' "
00:42.52kergothmikelp: heh.  perhaps read config.sub/config.guess to see what it supports for your arch?
00:43.10kergothwe replace those in every build, so you can be certain they're all the same
00:45.03mikelpooh, bitch, this would work if we didn't lower case BUILD_OS
00:45.16mikelpconfig.guess knows about Darwin, but not darwin
00:45.27kergothah, so you'll have to override that too then
00:45.45mikelpIs there a good reason why we do .lower it ?
00:45.49kergothin linux we have to, since uname shows Linux and configure expects linux
00:45.50kergoth(iirc)
00:45.57mikelpaha
00:46.11kergothnice and consistent, eh?
00:46.12kergothheh
00:46.13kergoth:\
00:46.40mikelpyep, my Suse boxen gives GNU/Linux
00:48.48zeckenite guys
00:48.53kergothnight zecke
00:48.56mikelplater dude
00:55.36emteshould have it parse Darwin into darwin so the case isnt an issue :)
00:56.05*** join/#oe dkey (~dkey@L0004P21.dipool.highway.telekom.at)
00:56.14kergothBUILD_OS_Darwin = "darwin"
00:56.16kergothheh.
00:56.35kergothactually, since BUILD_OS itself is in OVERRIDES, i wonder if thatd work
00:56.39emteactually you could do it in the oe  
00:56.46kergoth"the oe"?
00:56.50kergothevery .oe file?
00:57.01emteeither manually or with a python script
00:57.11kergothare you on crack?
00:57.14kergoththat doesnt make any sense
00:57.25emteno, i just woke up ... had a nap
00:57.32kergoththe gnu folks wouldnt accept a patch changing config.guess/config.sub, imho
00:57.33emte... your right it doesnt
00:57.40kergothfar better off correcting our input to it
00:58.11emtebut i did mean our input
00:58.26mikelppowerpc-darwin passes config.sub
00:58.34mikelpand ppc64-darwin
00:58.47kergothcool
00:59.17mikelpthe cp bug is annoying
00:59.32mikelpsomething is trying to copy in do_patch.....
01:00.34emtebtw ... eitehr oi dreamed it or  you guys mentioned sometime about trying to replace the python tools?
01:00.44emteeither i*
01:00.55mikelpyou dreamt it
01:01.02emteah
01:02.46*** join/#oe thatweasel (~thatwease@pool-68-237-211-181.ny325.east.verizon.net)
01:09.40mikelpkergoth: coreutils needs to make autoconf-native, which patches and hits the cp -a issue ;)
01:11.14kergothheh
01:11.15kergothdoh
01:13.53emtehmm should i finish fixing picogui-sdl or start preping my system for a reinstall ...
01:16.12mikelpcoreutils-native wouldn't help anyway
01:16.48mikelpsystem('/bin/cp', '-a', $fname, $backupname);
01:42.46*** join/#oe topic_ (~topic@p508AE396.dip.t-dialin.net)
01:52.39kergothmikelp: heh, you seem to be good at that
01:52.46*** join/#oe zwi (~zwi@216.88.131.43)
01:53.43mikelpkergoth: BUILD_LDFLAGS include gnu ld specific options ;)
01:54.49mikelpIts my own stupid fault to always wanting to try the most bizarre combinations
01:57.18mikelpsilly question, does SRC_URI_append_blah accept any definition of blah, or is it only target machine ?
01:57.59kergothit goes by OVERRIDES>
01:58.07kergothanything in OVERRIDES can be used as a conditional for a variable
01:58.09kergothread oe.conf
01:58.53mikelpooh nifty, its picks up TARGET_OS
01:59.33mikelpA quick add of BUILD_OS and I can fix a whole lot of these without screwing everything else up
02:00.18kergoththat was the idea :)
02:01.29mikelpHas anyone actually tried to use OE on a non gnu/cygwin OS ?
02:03.20kergothpeople have tested it on bsd and others
02:03.27kergothbut i dont think anyone's doing so actively
02:03.33kergothiirc there are some notes on the wiki
02:03.55mikelpnod
02:04.16mikelpHopefully, once all the -native stuff builds it'll be a lot easier
02:07.15mikelpnice, scummvm just failed on the opteron
02:15.19mikelp-L/home/phillim/zaurus/20041008/opie_tmp/staging/x86_64-linux/bin/../lib
02:15.23mikelpthat's not good
02:15.39*** join/#oe tigrux (~tigrux@201.129.245.218)
02:18.02W8TVIso is the SL-C3000 still running the system from the flash?
02:20.23*** join/#oe dkey (~dkey@L0004P04.dipool.highway.telekom.at)
02:27.23mikelp--with-mad-prefix=${STAGING_BINDIR}/..
02:29.15mikelpadd in a little    MAD_LIBS="-L$_prefix/lib from configure and you have nice -L native_libs in LDFLAGS
02:31.35mikelpwhich is all fine for mad, as no libmad exists in there, but libqte.a does
02:31.38mikelpand hence boom
02:31.42kergothheh
02:31.49kergoththats why i prefer configure scripts that let you set the include and lib dirs directly
02:31.56kergoth"prefix" assumes a specific structure
02:32.06mikelpIt works wrongly on 32bit
02:32.22mikelpIt links in the 32bit native lib, but doesn't fail because of it
02:32.35*** join/#oe minipanda (~zhanght@219.236.20.244)
02:33.31mikelpor when they use a app-config --libs
02:38.12*** join/#oe cterry (~chester_t@64-126-91-238-dhcp-kc.everestkc.net)
02:38.38cterryYo
02:52.49*** join/#oe slimspace|away (~darkstar6@navi.cs.colorado.edu)
02:56.16CIA-203phillim 07 * r1.2370 10packages/scummvm/scummvm_0.6.1b.oe: Incorrect mad-prefix was causing a native lib path to be included.
03:07.31emtehail scumm
03:22.39kergothheh
03:22.59kergothtaught my oevar class to expand itself, using %{}
03:23.00emte:)
03:23.06emte?
03:23.17kergothexperimenting with a replacement for the oe metadata handling.
03:23.28kergothsimplifies the syntax when using it greatly
03:24.16treke|homeha. the last qpe-gaim release was done on my birthday :p
03:24.22kergothheh
03:24.27mjcheh
03:25.44mikelpconfigure: error: source directory already configured; run "make distclean" there first
03:25.53kergothbug in autoconf
03:25.55mikelp~lart autotools
03:25.57kergothcant build in a dir with symlinks
03:26.15kergothfeel free to fix it, its one of those long standing annoying bugs nobody's been sufficiently annoyed to fix.
03:26.18kergothheh
03:26.19mikelpOooh, that's annoying
03:27.56mikelpnod. I'm stubborn enough to refuse to give in and move the files over
03:31.07emteis it furry with curly hair and an annoying bark?
03:32.07emte:P
03:33.00kergothmikelp: heh, wrote wrapper functions around this store i'm playing with, to make it compatible with the existing codebase.  gonna drop it in and see how it behaves :)
03:33.06kergoths/store/dict/
03:40.46*** join/#oe cdollar393 (~cdollar39@d3-240.rb.laj.centurytel.net)
03:41.02*** join/#oe cdollar393 (~cdollar39@d3-240.rb.laj.centurytel.net)
03:46.27treke|homekergoth: haha did you see the news about rexx?
03:46.40treke|homewasn't masticator talking about wanting to see that opened up?
03:47.32mikelpkergoth: crazy dude
03:49.23mikelpThis autoconf bug is annoying, exactly the same line of code in a test shell script doesn't do it
03:49.51treke|homeah mikelp, you'll do
03:50.01treke|homeis "chuffed" a good thing or bad thing?
03:50.08mikelpgood
03:50.13mikelpas in "well chuffed"
03:50.14kergothtreke|home: nope, didnt see the news, i'm outta the loop
03:50.21treke|homemikelp: gracias
03:50.42treke|homeyou make a fine enlgish to english translator
03:50.47treke|homemikelp: scummvm?
03:50.56treke|homewhat did I do to it?
03:51.07mikelpadded the .oe ;)
03:51.16treke|homeoh
03:51.35treke|homeI'll refrain from adding things to oe in the future
03:51.54kergoth..
03:52.10mikelpIt wasn't that bad
03:52.11*** join/#oe treke|home (~ggilbert@lsanca2-ar31-4-33-244-178.lsanca2.dsl-verizon.net)
03:52.40mikelptreke|home: I'm impressed it worked at all on 32bit, was finding native libs
03:52.54treke|homehehe
03:52.58treke|homethat surprises me
03:53.01treke|homeit shouldn't
03:53.31treke|homenot like it's autoconfed or any crap like that
03:53.53mikelp<PROTECTED>
04:00.47*** join/#oe dkey (~dkey@L0003P09.dipool.highway.telekom.at)
04:06.43emtebe nice when my book gets to std:: as opposed to using namespace std;
04:09.15*** join/#oe gate (~gate@dhcp160178111.columbus.rr.com)
04:09.59mikelpno bloody clue what's causing it though
04:18.51emtehmm
04:22.40CIA-203kergoth 07 * r1.792 10oe-newdata/bin/evildict.py:
04:22.40CIA-2evildict: Add expansion code to oevar, including %{} variable expansion and
04:22.40CIA-2backslash expansion (primarily for newlines).
04:39.07*** join/#oe lance (~charlyn@N453P006.adsl.highway.telekom.at)
04:44.01mikelpaha
04:44.06mikelpI think I know
04:59.13mikelpwish I could reproduce it outside of the configure script though
05:01.06*** join/#oe dkey (~dkey@L0003P23.dipool.highway.telekom.at)
05:14.48kergothheh, evildict is pretty fast
05:17.22CIA-203kergoth 07 * r1.793 10oe-newdata/bin/ (6 files in 3 dirs):
05:17.22CIA-2Add more wrappers around the new metadata handling to work with existing code,
05:17.22CIA-2and fix various spots around the oe core that were manually intializing the
05:17.22CIA-2data store rather than calling oe.data.init().
05:20.34mikelpmofo
05:23.09mikelpcall /bin/sh config.status from the cmd line works
05:23.21mikelpcalled from Make / ltmain.sh it fails
05:23.42kergothodd
05:24.23mikelpIt's all to do with the shell
05:34.44mikelpahahahaha
05:34.48mikelpgot the bastard
05:34.52mikelpIts us
05:35.57mikelpJust before do_compile is called in run_blah, we do a cd /blah
05:36.05mikelpbut our cd /blah is the real path
05:36.12mikelpmake it the symlink path and it works
05:36.50kergothhaha
05:36.51kergothoh man
05:38.37mikelpevery other path in run_ is the symlink path too
05:38.39CIA-203kergoth 07 * r1.794 10oe-newdata/bin/evildict.py: Teach oedict.update() to also be able to update from a normal dict, and fix the expand() comaptibility wrapper to handle expansion errors.
05:41.29*** join/#oe dkey (~dkey@L0004P30.dipool.highway.telekom.at)
05:41.54mikelpkergoth: what piece of code builds the run_ script ?
05:41.59mikelpsave me having to hunt it down
05:42.58kergothmikelp: oe.build.exec_func()
05:46.33mikelpf.write(cd %s\n" % os.getcwd()) looks like the bastard
05:47.07kergoththat doesnt look right.  each task has a flag already that has the directory it cd's to
05:47.17kergothwhich is where it should be going, not cwd
05:47.21kergothcourse i dont have the code in front of me
05:48.23mikelpthat line's from exec_func_shell
05:51.57mikelpos.getcwd definitely returns the real path, just tried in the interpreter
05:56.08CIA-203kergoth 07 * r1.795 10oe-newdata/bin/evildict.py: New metadata work: added a copyright header, and added an iterator for oevar's.
05:57.58CIA-203kergoth 07 * r1.796 10oe-newdata/bin/evildict.py: Make use of the oevar iterator instead of calling keys() all the time.
06:03.56CIA-203kergoth 07 * r1.797 10oe-newdata/bin/evildict.py: New metadata handling: unbork expansion of variables containing newlines.
06:32.14*** join/#oe dkey| (~dkey@L0005P30.dipool.highway.telekom.at)
06:41.10mikelp*sigh* I can't see an obvious answer to this in Python
06:45.47mikelpAh, screw it. the dictionary for the package knows where its meant to be, I'll get it from there
06:45.58mikelpbut that's something for tomorrow (or later today... ;)
07:08.13kergoth<PROTECTED>
07:08.24kergoth.7 seconds in an unknown execution of a string
07:08.27kergothand i have no idea where it is
07:08.27kergothhmm
07:35.26*** join/#oe ar_ (~ar@port-ip-213-211-233-133.reverse.mdcc-fun.de)
07:36.40*** join/#oe Disconnect (~dis@arent.you.special.you.gotontheinter.net)
07:43.36Pend-Goneheh 2.4M tmp area
07:43.43Pend-Gonethat's better than yesterday's 100k
07:44.38Pend-Goneautoconf-native booboo.. heh ;-) swweeet
08:57.42*** join/#oe |dkey| (~dkey@L0009P27.dipool.highway.telekom.at)
09:26.43*** join/#oe Cwiiis (~cwiiis@user-214-207-151-83.e7even.com)
09:38.59*** join/#oe jacques_ (~jacques@c-67-171-198-193.client.comcast.net)
09:40.58CIA-203kergoth 07 * r1.798 10oe-newdata/bin/evildict.py:
09:40.58CIA-2bin/evildict.py:
09:40.58CIA-2<PROTECTED>
09:40.58CIA-2<PROTECTED>
09:51.56*** join/#oe cedric (~cedric@voulx.bluebugs.org)
09:53.28*** join/#oe Hertog (~gertjan@cc233190-a.ensch1.ov.home.nl)
10:03.33*** join/#oe darkschneider (~gab@81-208-36-80.fastres.net)
10:22.11CIA-203mickeyl 07 * r1.2368 10packages/python/python-pyqwt_3.10.oe: PyQWT: catch up w/ qt staging changes and convert to new sip build scheme
10:22.11CIA-203mickeyl 07 * r1.2364.1.1 10packages/python/ (python-2.3.4-manifest.inc python_2.3.4.oe): eliminate bogus , in python subpackage dependencies
10:23.15kergothhey mickeyl
10:27.55mickeylmorning kergoth
10:28.00mickeyl(local time, that is)
10:28.08mickeylisn'it pretty in the middle of the nite for you?
10:28.14kergothyep, 5am
10:28.19mickeyld'oh :)
10:28.59kergothhad a bit of fun playing with a metadata store thats easier to use
10:29.01kergoth<PROTECTED>
10:29.01kergoth<PROTECTED>
10:29.01kergoth<PROTECTED>
10:29.01kergoth<PROTECTED>
10:29.07kergothheh.
10:30.22mickeylah nice, a convenient replacement for the get/set var things
10:30.29kergothyep
10:30.54kergothit has slightly more overhead due to all the reimplemented getattr/setattr/etc, but its not bad.  been profiling and optimizing
10:31.05kergothwrote wrappers that implement the old api, its a drop in replacement
10:31.12kergoth(well, not including overrides)
10:31.35mickeylnice. btw., how's the state of the COW stuff? how far away is it from being usable?
10:32.03kergothseb's on like run 5, apparently only a couple bugs left to iron out
10:39.01*** join/#oe [jmau] (~jmau@pD9E5DCB3.dip.t-dialin.net)
10:41.00CIA-203mickeyl 07 * r1.2370 10packages/meta/task-python-everything_0.9.0.oe: fix typo in task-python-everything
11:00.21CIA-203br1 07 * r1.2366.2.1 10packages/rp-pppoe/rp-pppoe_3.5.oe: rp-pppoe has changed it's download url
11:02.45*** join/#oe treke|ho1e (~ggilbert@lsanca2-ar31-4-33-244-178.lsanca2.dsl-verizon.net)
11:13.55CIA-203ral 07 * r1.2372 10packages/opie-eye/opie-eye_1.1.6.oe: fixed opie-eye installer
11:16.03*** join/#oe zecke (~ich@83.169.144.140)
11:17.49*** join/#oe |dkey| (~dkey@L0007P16.dipool.highway.telekom.at)
11:41.01*** join/#oe pb_ (~pb@cpc2-cmbg6-5-0-cust48.cmbg.cable.ntl.com)
11:43.51CIA-203pb 07 * r1.2366.3.1 10packages/nsqld/nsqld_0.5.3.oe: add nsqld, courtesy Tim Teulings
11:44.24*** join/#oe Azile_ (~amc@dsl-62-3-114-252.zen.co.uk)
11:50.40*** join/#oe qfh (~qfh@ppp-82-135-12-155.mnet-online.de)
11:54.04*** join/#oe |dkey| (~dkey@L0003P19.dipool.highway.telekom.at)
12:03.45koenpb_: depmod give me a Q_MODULES not supported in my OE image
12:04.17koencould it be modutils collateral doesn't work properly and it nees module init-tools?
12:04.23koen(ipaq-pxa-2.6 btw)
12:04.57pb_yes
12:05.16pb_you need to use the depmod from module-init-tools, not modutils-depmod.
12:05.25pb_(modutils-collateral probably isn't directly relevant)
12:06.02koeni'll add that to ipaq-pxa-2.6 RDEPENDS somewehre
12:06.10pb_ok, good idea
12:06.56koenis it safe to remove modutils-collateral?
12:09.11zeckekoen: as long as your kernel.oe  has lines like autoload_irda = "..." it is ok to remove modutils-collateral
12:11.30koenzecke: i'll leave it in for now
12:12.15koenpb_: could you add the module-init-tools to BOOTSTRAP_RDEPENDS in ipaq-pxa-2.6.conf in bk?
12:21.28pb_ok
12:31.18pb_hi zecke
12:32.14CIA-203pb 07 * r1.2374 10packages/conf/machine/ipaq-pxa-2.6.conf:
12:32.14CIA-2add module-init-tools to {R}DEPENDS.
12:32.14CIA-2requested by Koen Kooi
12:46.45*** join/#oe lance (~charlyn@N647P012.adsl.highway.telekom.at)