/srv/irclogs.ubuntu.com/2005/04/26/#ubuntu-kernel.txt

cartel_hppa is the superdome architecture isnt it?12:01
T-Noneamong other things12:02
cartel_wildcat?12:02
=== zul [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel
zulhey12:32
cartel_im having troubles unpatching the kernel-source12:56
zulunpatching? like how12:56
cartel_patches are silent12:56
cartel_untar linux-source-2.6.10 and run unpatch/debian, it thinks it is not a debian kernel tree..12:57
zulcartel_: is there a debian directory?12:57
cartel_yes12:57
zulcartel_,: do this fakeroot debian/rules unpack in the main kernel directory if you want to unpack it and patch it12:58
cartel_i want to apply a top level patch before the debian patches12:58
zulwhich version -33?12:59
cartel_2.6.10-3412:59
zulthen you have to make the patch into a dpatch and stick the name of the patch in debian/patches/00List-3401:00
cartel_no way01:00
cartel_dpatch???01:01
zulyes way01:01
zulyep01:01
cartel_you mean so it appears in /usr/src/kernel-patches ?01:01
zullook at one of the patches in debian/patches for an example01:01
cartel_ive already done that01:01
cartel_now for some reason linux-source-2.6.10 has all the patches applied but is missing debian/ folder01:02
zulapt-get source linux-source-2.6.10 right?01:02
cartel_yea01:03
cartel_no, apt-get source linux-image-2.6.10-5-k701:03
cartel_which gets linux-source-2.6.1001:03
zulthen there suppose to be a debian directory in linux-source-2.6.10 directory correct?01:03
cartel_but its bung01:03
cartel_yeah01:03
zuland in the directory there is a debian directory correct with the rules file and stuff01:03
cartel_nope01:04
zuldo apt-get source linux-source-2.6.10 and there should be one in there01:04
cartel_debian/ not there01:04
cartel_ok, now i have debian/01:05
zulhold on a sec01:05
cartel_# /usr/src/linux-patches/i386/2.6.10/unpatch/debian01:05
cartel_No version.Debian file, assuming pristine Linux 2.6.1001:05
cartel_bung01:06
cartel_what should version.Debian be?01:07
zulhttp://www.ubuntulinux.org/wiki/KernelBuildpackageDetailedHowto/01:07
cartel_why change name from kernel-source to linux-source? seems pointless01:07
zulthat is the way its done01:08
cartel_what package is dch in?01:08
cartel_whats the point?01:08
cartel_its a stream edit on the debian rules... why??01:09
cartel_these instructions afre for warty01:09
zulbecause the rules file reads the changelog file01:09
cartel_do i still need restricted?01:09
zulthey are still valid for hoary just change the version numbers01:09
zulyou shouldnt01:09
cartel_where is dch?01:09
zulcartel_, check #ubuntu01:10
zulor do an apt-get search01:10
cartel_apt-cache search01:10
cartel_how do i know a file is an dpatch as opposed to a normal patch?01:14
zulbecause its has a @DPATCH@ in it01:15
cartel_well i hate to dissapoint but not even the kernel-patch has that01:16
zulyou didnt read the wiki did you?01:16
cartel_so this doc is worthless because kernel maintainers dont follow their own guidelines01:16
zulcartel_ this isnt a support channel try #ubuntu 01:17
cartel_im trying to build a xen-linux-image-2.6.10 for the community01:17
zulgood luck..01:18
cartel_im not looking for support im looking for the maintainers not a noob wiki page01:18
zulyou might want to talk to to schweeb in #ubuntu-motu01:18
cartel_i dont understand why so many deltas from debian policy01:18
cartel_and why kernel-source-2.6.10 is so cornholed01:18
cartel_ok thx01:18
cartel_it just makes things real confusing01:21
T-Nonei'm just hopping by, but are you trying to *unpatch* a freshly unpacked kernel source tree?01:30
=== zul [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel
lamontmoo02:27
zulhey lamont 02:28
zulhow goes it?02:28
lamontzul: taxingly02:28
zulheh..i already did mine02:28
=== T-None groans at lamont :)
lamontzul: finally got the last of the info I need this morning.02:29
zulcool..so you werent slacking ;)02:29
lamontwell, stalling, yes.  slacking, no.02:30
lamontanyway,  not paying too much attention over here for the next while...02:30
zulokie dokie02:31
zulim taking a break02:55
zulbbl02:55
=== lamont completes pass 1, debates hurting his wife
zulheh...how much are you going to be owing?04:46
lamontzul: other way around.  been waiting for her to give me info... we get a pretty nice chunk back04:48
zulcool.04:48
zullater...05:12
=== zul [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has left #ubuntu-kernel ["Leaving"]
fabbionemorning06:04
T-Nonemoo-rning11:21
fabbionehey T11:21
=== T-None is now known as T-Bone
T-Bonefabbione: you're on the Debian apache team, or am I confused with some other Italian guy?11:24
fabbionethat would be me.. yes11:24
T-Boneok11:24
T-Bonequick offtopic question if you don't mind then :)11:24
fabbionego ahead11:24
T-Bonei just received a wishlist bugreport for an apache module i both write and maintain. That's a pretty indexing module, so to speak, and it uses some icons/css to format the output. When i packaged the module, i put these files into /var/www/modulename, because i thought this was the correct policy. The user wants them in /usr/share/modulename. Who's right? :)11:26
fabbionethe user11:27
fabbioneyou have no rights to write in /var/www/11:27
fabbione+ if you write stuff in there, you need to treat them as config files11:27
T-Bonegrumble11:28
T-Bonethat module has been in for about 2 years now, and I wonder whether it'd be a good thing to break it before sarge release :P11:29
T-Bonei suppose i'll change that post-sarge then11:29
fabbioneT-Bone: that could be considered RC bug11:30
fabbionesince /var/www contains user data11:30
T-Boneno11:30
T-Boneerr11:30
fabbioneand you can mangle them just adding your stuff11:30
T-Boneah11:30
T-Bonehmm11:30
T-Boneshit11:31
T-Bonehow comes this passed NEW :P11:31
Mithrandiradd a symlink on upgrades11:31
Mithrandir(if the directory exists)11:31
Mithrandir(and remove it in postrm, if you added it on upgrade)11:31
T-Boneerr11:31
fabbionehmmm11:31
fabbionei don't think that would work.. what if i have my own /var/www/$foo/$bar.html11:32
fabbionelater i install T-crap module11:32
fabbioneand he moves stuff around with a symlink killing my bar.html?11:32
fabbionei would lart T-crap to death11:32
T-Bonesigh11:32
fabbionehe will need to check the contents of what he created11:33
fabbioneas i said.. treat them as config files11:33
fabbionefor new install move them to /usr/share11:33
fabbionefor upgrades.. you are doomed11:33
T-Bone:(11:33
Mithrandireither NEWS.Debian or debconf note (preferably the former) to tell the user the files have moved.11:34
fabbioneT-Bone: that's a really bad situation.. or rather technically interesting to handle properly11:34
T-Bonewhat if i detect the existence of /var/www/foo, and prompts the user an alert?11:34
fabbioneT-Bone: seems the right thing to do11:35
=== T-Bone will need to improve his pre* post* skills, btw :P
fabbionebut also add a very detailed README on how to do the right thing11:36
T-Bonehmmm11:36
T-Bonelemme think11:36
T-Bonewhat if i detect /var/www/foo and then 1) not install /usr/share/foo, 2) move /var/www/foo to /usr/share/foo ?11:37
T-Bone(hoping it's possible to do that)11:37
fabbionenpe11:37
fabbionethat won't work11:37
T-Bonewhy?11:37
fabbioneif you detect /var/www/foo you warn that stuff has moved to /usr/share/foo11:37
fabbionedo you create /var/www/foo dinamycally?11:38
fabbioneor is it shipped in the package?11:38
T-Bonewell actually one of the reasons why it has been placed in /var/www/foo was that the user was supposedly empowered to add his own CSS files11:38
T-Boneshipped11:38
fabbionewell you will have to ship it somewhere.. and that somwehre has to be /usr/share11:39
fabbionethe problem is that upgrading, dpkg will remove uncoditionally what is in /var/www/foo11:39
T-Boneand what about the user that wants to add his own cruft? He should put it in /usr/share/foo?11:39
T-Boneugh11:40
fabbioneT-Bone: no, you need to suggest somewhere to copy or symlink11:40
T-Bonemy oh my11:40
T-Bonewhy make things simple when we can complicate them to death, heh? :P11:40
fabbioneT-Bone: YOU ASKED FOR IT: BECAUSE YOU ARE FRENCH!11:40
=== fabbione ducks
T-Bonehmm11:40
T-Bonebtw, why would dpkg remove inconditionally. When it finds foreign files in a location, it lets the directory in place, unless you --purge11:41
T-Boneiirc11:41
fabbioneyes.. the foreign files...11:41
fabbionewhat about files you ship that have been customized?11:42
T-Bonethe user shouldn't do that :)11:42
T-Boneit's explicitely detailed in docs11:42
fabbionethe user can do whatever he wants with his data11:42
fabbioneand /var/www is data11:42
T-Bonesigh11:42
=== T-Bone contemplates breaking the users that don't read docs
fabbioneRC bug11:43
fabbioneuser will break you if you touch his data11:43
T-Bonethat's why i won't change it before sarge11:43
T-Boneconsidering the limited audience (according to popcon) of the module, the fact that it's a 0.xx (deemed beta in the docs) version, well11:44
fabbioneT-Bone: do something sane...11:44
fabbioneport it to apache211:45
T-Boneit is11:45
fabbioneand drop the a1.3 package11:45
fabbione:)11:45
T-Bonethere are both11:45
T-Boneand i don't see how that fix my problem11:45
fabbioneoh god11:45
T-Bone(both built out of same source, btw)11:45
fabbionewell i though perhaps you were not insane enough to mangle a2 as well11:45
T-Bonethe idea is that both package share the same common files11:45
T-Bonethe user moving from apache1.3 to apache2 has nothing to do (almost)11:46
T-Bonei hate murphy's law11:47
=== T-Bone has NFC how to "do something sane" :(
fabbioneT-Bone: BECAUSE YOU ARE FRENCH! <- /me repeats11:49
T-Bonefabbione: and btw, *NEVER* underestimate my insanity11:49
T-Bonethis is not helping, you italian punk11:49
T-Bone:P11:49
=== T-Bone thinks breakfast will help considering this issue
=== T-Bone is now known as T-Gone
T-Gonemjg59: sorry for the lag in replying. Forgot to mention again i'm not subscribed to d-project :)02:11
=== T-Gone -> food
=== zul [~chuck@198.62.158.205] has joined #ubuntu-kernel
zulhola02:51
jbaileyHeya Chuck02:55
jbaileylamont: ping?02:56
fabbionehi zul02:58
zulhey fabbione 02:59
zuldie!03:28
zulcool..speakers has been announced for OLS http://www.linuxsymposium.org/2005/speakers.php?types=TALK04:45
dilingerdo they stream that?04:47
=== dilinger would love to see james bottomley and len brown speak
kyle_no, but the audio is recorded.04:48
zulor you could just go to ottawa like everyone else04:51
zulyou bum :)04:51
kyle_indeed.04:52
dilingerhehe04:53
dilingeryea, i'm thinking about it04:53
zulcool...then we can go party04:53
kyle_word04:54
zulof course if im working and if i have money 04:54
dilingeryea, hopefully i'll know whether i'll be living in colorado or nyc soon04:56
dilingersince if i'm in nyc, going to OLS is not a problem..04:56
zulnyc is a bit more closer 04:56
=== fabbione yawns
zulhey fabbione 04:57
fabbionezul: time to merge from your branch...04:57
zulgo ahead..04:57
kyle_morning fabbione04:57
fabbionezul: can i lart you a bit first?04:57
zulno you cant04:58
zul:)04:58
fabbionehey kyle_ .. it's late afternoon.. but goodmorning to you to04:58
zulfabbione: er i mean go right ahead04:58
kyle_gah, why have i got a tail.04:58
=== kyle_ is now known as kylem
fabbionezul: is your archive online?04:59
zulyep04:59
fabbioneoh i see.. just slow to death04:59
zulyeah04:59
fabbionedid you compile the new patches?04:59
fabbioneor are they totally untested?04:59
zulno i havent had a chance yet04:59
=== fabbione sighs
kylemwow. git could really use a friendlier interface.04:59
fabbionei haven't touched it yet05:00
zulme either05:01
zulfabbione: so let me have it05:01
fabbionezul: please test that at least the patches build05:01
fabbionethan i will merge05:02
fabbionealso.. it is a good idea for you to merge regularly from pre105:02
fabbioneso that there are not too many differences in the archive when we push/pull stuff05:02
zulfabbione: i usually do test the patches i just havent had time to05:03
zultoday at least05:05
zulthats weird /bin/sh: line 0: [: /usr/src/kernel-patches/i386: binary operator expected05:07
fabbionewhat did you break?05:08
zulnothing as far as i know...its breezy though05:08
fabbioneprobably bash is borked05:08
zulprobably05:08
fabbionewhat operation did you try?05:10
zuldpkg-buildpackage05:10
fabbionetry fakeroot make -f debian/rules build05:13
zulok ill try that05:13
zulneed to fix a typo first05:13
fabbioneworks here.. but i didn't update since this morning05:14
zuli updated about an hour ago05:14
zulbuilding away...build bluild build05:35
zulbbl...lunch05:46
zulbrb07:13
fabbionezul: did it build?07:41
=== zul [~chuck@198.62.158.205] has joined #ubuntu-kernel
fabbionezul: did it build?08:01
zulits still building08:01
zulfabbione: do you want me to take care of unionfs later as well?08:07
fabbionezul: i would like to merge from you first08:08
fabbionethan we can take a look :)08:08
zulij08:08
zuldoh..ok08:08
zulgah...im getting autoconf.h errors now08:08
zulinclude/linux/autoconf.h:1419:19: warning: ISO C requires whitespace after the macro name08:08
fabbioneare you building with gcc-4?08:09
zulnope its gcc-3.308:09
zulas in the makefile08:09
fabbioneit builded here without any problem08:09
zuli did an update this morning maybe something went foobared with kernel-headers08:10
zuljbailey: ping08:10
fabbionethe kernel doesn't use kernel-headers to build08:10
zulokie dokie :)08:10
zulthen maybe its me08:10
jbaileyzul: 'sup?08:10
zuljbailey: unping :)08:10
zulsorry08:11
zuli sucketh this week08:11
zulmore so than usual08:13
jbaileyThat's okay.  People pay lots of money fo...  err..  08:13
jbaileyNM08:13
zuloh i know08:13
fabbioneAOE is broken with devfs08:25
fabbionewe need to kill DEVFS!08:25
fabbioneanyway08:25
fabbionedinner time08:25
=== fabbione kills DEVFS
zulproblems?08:59
fabbionezul: yes.. it's crap :)09:04
Mithrandirpft09:05
Mithrandirdo we have some CotD packages of 2.6.12?09:05
zuldilinger: heh...read your blog09:06
zul"Sometimes when I'm alone, I google myself."09:06
fabbioneahha09:06
fabbioneactually it would be very simple to fix aoe to support devfs09:07
fabbionebut do we care?09:07
zulwhat the hell is aoe?09:07
kylemata over ethernet09:07
fabbioneATA over Ethernet09:07
Mithrandirwe should be able to turn off devfs now, I'd imagine09:07
kylemaka: puke09:07
zulah..09:07
fabbionekylem: well it's mainline.. so it can't be THAT bad09:07
kylemlots of things are mainline and fucking digusting. :)09:08
fabbioneMithrandir: yeah.. i think d-i is no devfs safe09:08
kylemlook in drivers/net someday :)09:08
fabbionekylem: well i am checking AOE and GNDB atm09:08
Mithrandirfabbione: d-i uses udev in devfs mode.09:08
zulwell turn if off and if people complain ;)09:08
fabbionekylem: i did :(09:08
fabbionezul: devfs is used only by the installer right now09:08
fabbioneso i think we could disable it and let Kamion deal with it :P09:09
fabbioneyou know.. ain't my problem...09:09
fabbionewe did NOTTING09:09
kylemfabbione, my sympathies ;_)09:09
fabbioneright?09:09
zuli say go for it09:09
fabbionei want to talk with Kamion first09:10
fabbioneat least warn him09:10
zul...but im not kamion09:10
fabbioneand since he just left for LCA09:10
zuloh yeah...udu is coming up pretty quickly09:10
fabbionei am going to test both solutions...09:11
fabbionewithout devfs and fixing aoe :)09:11
fabbionei love challenges :)09:11
Mithrandirfabbione: you are insane :)09:12
fabbioneMithrandir: yes i know!09:13
fabbioneMithrandir: fixing aoe means adding 4 lines of code in the right place, one of which is an include09:13
fabbioneone a simple definition09:13
fabbioneand 2 to notify devfs of device add/remove09:14
fabbioneremoving the config means changing 1 line in approx 12 files...09:14
fabbionetheroretically fixing aoe is less intrusive :P09:14
Mithrandirfabbione: ok, cool enough.09:16
zulcrap it doesnt build09:18
zulgrrr..09:24
fabbionezul: error?09:31
zulyeah...ill fix it09:31
fabbionezul: just tell me the error :)09:31
zulmissing a header file09:31
fabbioneuhuh09:34
fabbionebooting with no devfs spits out a bunch of errors09:34
=== caldwell [~caldwell@cerberus.dtn.radian.biz] has joined #ubuntu-kernel
fabbionehmmmm09:45
zulcrap i need to head home09:47
zulill have this fixed when i get home tonight09:48
zullater09:48
zulfabbione: ill send you an email when its fixed09:49
fabbionethis starts to be interesting :)09:55
jbaileyfabbione: This -> NULLPOINTER Exception.  Clarify and resubmit?10:39
fabbionejbailey: uh?10:39
jbaileyfabbione: *exactly*10:39
fabbionejbailey: what that come from?10:40
jbailey<fabbione> this starts to be interesting :)10:40
jbaileyI'm wondering what you're talking about.10:40
fabbioneAHHHH10:40
fabbionethe AOE driver :)10:40
fabbioneeither the vserver is broken or the driver is :)10:41
fabbionebecuase they talk to eachother at init time10:41
fabbionebut they stop there10:41
fabbioneand it looks to me that the kernel driver doesn't tell the kernel properly that there is a new device10:41
jbaileyWhat's AOE?10:42
fabbioneAta over Ethernet10:44
jbaileyOh creepy.10:45
fabbioneyeah10:49
fabbionebut i was curious to see how it works10:49
fabbionethe client/server setup is very simple10:49
fabbionethat's why10:49
fabbioneyou run a daemon with 2 options10:49
fabbioneand modprobe on the client10:49
kylemit's probably better than NFS, heh.10:49
fabbionezack you share a block device10:50
fabbionekylem: that's for sure10:50
fabbionenow i am testing gndb10:50
fabbionebut it's giving the creep to compile the userland10:50
Mithrandirhm11:18
Mithrandirwhy does a build of linux-source-2.6.10-37 fail with :11:18
Mithrandirdrivers/acpi/ec.c: In function 'acpi_ec_gpe_query':11:18
Mithrandirdrivers/acpi/ec.c:519: error: label at end of compound statement11:18
Mithrandirm?11:18
fabbionegcc-411:22
Mithrandirhoooray!11:23
Mithrandir*sigh*11:23
=== Mithrandir redoes with CC set to gcc-3.3
lamontjbailey: ack11:38
jbaileylamont: Do you have a view of the build database that can show which packages failed that show up as uploaded fine in Debian?11:39
jbaileylamont: I'm interested in seeing which packages FTBFS based on toolchain changes, and that seems a likely way of seeing them.11:39
lamonthttp://people.ubuntu.com/buildLogs/Lists/ has a few files in it...11:39
lamontbreezy.failed.$arch is probably what you're after11:40
lamontof course, that says nothing about how they do in debian/11:40
jbaileyYeah, but it might give a start.11:40
jbaileyShould that be ~lamont ?11:40
jbaileyYup11:41
lamontdoh11:41
lamontyeah - ~lamont/buildLogs11:41
lamontmy b ad11:41
jbaileyIs "not ours" the same as quinn-diff's not-for-us ?11:42
lamontjbailey: kinda...11:44
lamontinstead of using --not-for-us, I use --failed -m 'not ours'11:45
jbaileySo these show no FTBFSs? =)11:46
jbaileyMakes me feel very good about the toolchain update. =)11:46
=== mjc [~mjc@c-66-176-195-87.hsd1.fl.comcast.net] has joined #ubuntu-kernel

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!