**** BEGIN LOGGING AT Sat Oct 30 02:59:57 2010 Oct 30 03:07:55 ping xMff Oct 30 03:08:01 here Oct 30 03:09:23 Hi. Just thought I'd let you know that I'm not going to be able to work on the sysupgrade+extroot stuff this weekend...I'm not sure how much time I will have through the week either. I'll try and make time, but I've got some major work to do right now. Oct 30 03:15:02 no problem Oct 30 03:15:08 it works well atm Oct 30 03:16:09 except i think the fsck of the rootfs doesn't work with modularized extroot....I plan on fixing that too....but that's really a minor issue IMO Oct 30 03:16:44 or rather modularized fsck Oct 30 03:17:53 hmm... somehow Firefox manages to slow luci down Oct 30 03:18:20 just tried it in chrome and everything is instantaneous Oct 30 03:18:33 xMff: yeah, I just noticed that too Oct 30 03:19:42 I even installed Firefox 4 in the hope it gets faster... Oct 30 03:20:08 xMff: with non-compiled lua it's really, really slow on Firefox...haven't tried uncompiled lua on Chromium (the basis of Chrome), but compiled lua on FF vs compiled lua on C is huge Oct 30 03:21:03 but I don't get why, its simple, clean html, very few js Oct 30 03:21:18 FF sucks? Oct 30 03:21:44 Seriously, I think it's the rendering engine Oct 30 03:21:55 they've improved FF, but it's still slow Oct 30 03:22:02 I have the feeling it has strange tcp handling as well Oct 30 03:22:15 hmmmm.....now that's an interesting possilbiity Oct 30 03:22:32 noticed weird behaviour when working on uhttpd Oct 30 03:22:56 I should probably get rid of the artwork too Oct 30 03:23:01 I've switch browsers because as soon I tried Chromium I realized there was a option besides FF (well Iceweasel) Oct 30 03:23:15 why? isn't that what the light theme is for? Oct 30 03:23:34 yeah but the one with images is the default and it slows down rendering a lot Oct 30 03:23:46 ah, maybe just make the light the default Oct 30 03:23:48 at the beginning we had it fixed Oct 30 03:24:02 that totally killed the performance on macbooks... heh Oct 30 03:24:14 heh Oct 30 03:24:30 hm, must rework the menu too Oct 30 03:24:53 its not iphone friendly Oct 30 03:25:16 because of the disappear when you scroll over, or the js or css? Oct 30 03:25:30 well, the touchfoo has no concept of mouseover Oct 30 03:25:46 and to get into a submenu you have to click, which also triggers a page reload Oct 30 03:25:56 I see, yeah, that would be problem Oct 30 03:26:52 can you do clickable menus without js (and not a page reload either)? Oct 30 03:27:01 or will you need js? Oct 30 03:27:08 I think I just make tabs Oct 30 03:27:30 I gave up on the no js policy, allowed some very slick ui concepts Oct 30 03:27:54 do you have a degradation to no js but more page reloads for the unlucky users? Oct 30 03:28:03 not anymore Oct 30 03:28:29 its too much work Oct 30 03:28:34 and it bloats the code a lot Oct 30 03:28:45 hmmmm.....well it's not like I've got the time to do it, so I won't complain, but it's a pity Oct 30 03:28:47 I currently try to make it smaller Oct 30 03:29:13 yeah, there is that Oct 30 03:29:24 it's not like consumer routers have non-js interfaces either Oct 30 03:29:34 lately even flash Oct 30 03:29:45 GrrhGhkk! Ugh! Oct 30 03:30:36 I saw it on some of the newer asus routers Oct 30 03:30:45 they used it to visualize the network topology Oct 30 03:31:02 I think in a few years they'll use canvas or svg instead Oct 30 03:31:10 I hate Flash, not the least because you have no choice but to use Adobe's player for most videos, and they still don't have the 64-bit latest player Oct 30 03:32:39 true Oct 30 03:33:21 I've heard HTML5 being touted as a solution...I hope it pans out Oct 30 03:39:00 I think it will Oct 30 03:58:37 i just used a 64-bit flash player on linux from adobe labs. Oct 30 04:03:21 tho they have nothing for non-intel platforms, which i use most :( Oct 30 04:25:31 ~. Oct 30 04:28:07 neutronscott: version 10? that's what sites are requiring now... Oct 30 04:28:29 yeah i mean youtube worked. Oct 30 04:29:16 oh, I see they have a preview release now Oct 30 04:30:24 i was amazed. :D Oct 30 04:36:34 omg the latency, trying to hit router upstairs. turned off wifi on wrt54g-tm and yep, mmc_over_gpio works now. Oct 30 04:36:49 guess i get to disable something in that driver Oct 30 07:05:46 build #21 of at91 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/21 Oct 30 08:18:18 build #19 of s3c24xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/s3c24xx/builds/19 Oct 30 10:05:03 acoul * r23718 /trunk/target/linux/generic/patches-2.6.36/512-yaffs_symlink_bug.patch: linux/generic: fix "rm: can't remove 'xyz': Directory not empty" symbolic link and directory issues on RouterBoard family devices. Oct 30 12:21:30 dingo * r23719 /packages/net/coova-chilli/Makefile: [patch-team] update coova-chilli to version 1.2.5 Signed by: hanno.schupp@gmail.com Oct 30 12:25:09 dingo * r23720 /packages/net/coova-chilli/patches/001-readd-macauth.patch: [patch-team] remove net/coova-chilli/patches/001-readd-macauth.patch Signed by: hanno.schupp@gmail.com Oct 30 15:27:35 mb__: why did you put the package-python.mk install into Host/Install? Oct 30 15:28:32 because it contains the path to host python. Oct 30 15:29:02 It's not 100% correct as-is, because hostpython is only installed in target python's install routine Oct 30 15:29:09 But I didn't want to change that for now. Oct 30 15:34:58 well, one problem is that if you change your target Host/Install is not rerun and you end up with no package-python.mk in $(STAGING_DIR)/mk thus all packages using it fail to build Oct 30 15:35:28 I guess you can move it back Oct 30 15:35:43 Somebody suggested (I forgot who) that I move it there Oct 30 15:35:59 In practice, if you need hostpython, you must depend on target python anyway. Oct 30 15:36:10 build #18 of adm5120 is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/adm5120/builds/18 Oct 30 15:36:19 (Yeah, that should not be the case, but it's like that for now) Oct 30 15:37:32 IMO we should not support switching targets without distclean anyway. But that's another topic. Oct 30 15:37:50 It just causes so many problems that I'm gonna do a dirclean anyway. Oct 30 16:14:01 build #17 of ps3 is complete: Failure [failed compile_6] Build details are at http://tksite.gotdns.org:8010/builders/ps3/builds/17 Oct 30 19:29:57 acinonyx * r23721 /packages/net/iodine/Makefile: [packages] iodine: Cleanup package Makefile Oct 30 19:30:06 acinonyx * r23722 /packages/net/quagga/patches/100-quagga_multipath_caching_policy.patch: [packages] quagga: Remove multipath caching policy patch since it is unsupported in recent kernels (#8136) Oct 30 19:35:40 seeing the following build breakage: http://pastebin.ca/1977500 -- I'm at SVN 23717. Oct 30 19:36:05 not sure why eglibc-2.12 is breaking. Oct 30 19:41:01 mb * r23723 /packages/Xorg/app/gpxview/: Remove gpxview Oct 30 19:41:16 mb * r23724 /feeds/desktop/apps/gpxview/ (6 files in 2 dirs): Add gpxview Oct 30 20:48:28 larsc: I'll check if I can fix the python stuff by completing the split-up. Oct 30 20:49:35 ok Oct 30 20:50:55 larsc: Not sure if it's possible to fix that, though. Oct 30 20:51:19 I'd like to keep the new HostPython macro. Where to put it? Oct 30 20:51:26 We don't have host-mk-files. Oct 30 20:52:21 So I'll probably end up moving the mk copy to target again. Oct 30 20:52:44 And simply require packages do a PKG_BUILD_DEPENDS:=python, even if they only need python/host. Oct 30 20:53:23 well, imo it would be fine to install the mk file into $(STAGING_DIR_HOST)/mk and create link to it in $(STAGING_DIR)/mk Oct 30 20:54:06 hm, this is a bit hairy. Oct 30 20:54:21 I'm not sure I want to go down that path Oct 30 20:54:32 all those inter-dependencies. Oct 30 20:54:57 the target python needs host python anyway Oct 30 20:55:05 Yeah right. Oct 30 20:55:14 But there are packages with only need host python Oct 30 20:55:15 so it should not be a problem imo Oct 30 20:55:27 well, thats no problem either Oct 30 20:55:52 yes, a PKG_BUILD_DEPENDS:=python should be fine there, too Oct 30 20:56:05 but those should use the mk from the host staging dir Oct 30 20:56:36 why? Oct 30 20:57:13 because if they only depend on host python, the target python might not be installed Oct 30 20:57:35 and host packages shouldn't put any files in the target staging_dir Oct 30 20:57:42 There's one thing I don't understand. That is the "hostpython" binary that is copied to target staging dir. It does some magic that I don't get. Oct 30 20:57:58 what kind of magic? Oct 30 20:57:59 larsc: Well, they need to depend on target python Oct 30 20:58:11 Well, I do not know. That's the problem. :) Oct 30 20:58:31 It does some search path magic or something. See the comment in the makefile, too. Oct 30 20:58:52 it is related to setup.py scripts Oct 30 20:59:10 if you use STAGING_DIR_HOST/.../python for setup.py, it fails. Oct 30 21:00:27 So, you see, this is a complete mess. And this is why I'll most likely just move the .mk copying and don't touch it anymore :) Oct 30 21:01:08 ok Oct 30 21:01:16 setup.py seems to have horrible crosscompile semantics for C modules Oct 30 21:01:29 yea Oct 30 21:01:35 it uses some magic to detect things Oct 30 21:03:36 I'll move the .mk copying and commit that. Afterwards I will _try_ to check if I can fix the rest. Oct 30 21:10:33 Is eglibc broken? Oct 30 21:12:25 compiling build_dir/toolchain-i386_gcc-4.5.1_eglibc-2.12/eglibc-2.12-r10495-final/nscd/nscd is failing. Oct 30 21:19:06 libssp Oct 30 21:21:26 philipp64|laptop: maybe have configure --disable-libssp ? Oct 30 21:25:09 mb * r23725 /packages/lang/python/Makefile: python: Move .mk copying to InstallDev Oct 30 21:43:30 neutronscott: sorry, where? Oct 30 22:20:02 hello, we are just working on a Samsung SMT-G3200, with an infineon Amazon SoC, unfortunately the PCI driver is not really working correctly, so that the WLAN card is not recognized correctly (the card is recognized by lspci but not by the driver, it is an Atheros AR5008, with ath9k), what can we do? Oct 30 22:20:35 hlu: probably related to the EBU swapping configuration Oct 30 22:20:50 hlu: the same problem existed for danube in the ifxmips target at some point Oct 30 22:20:53 but it was fixed there Oct 30 22:21:02 blogic will know more about this Oct 30 22:22:55 build #15 of xburst is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/xburst/builds/15 Oct 30 22:23:25 nbd: do you know by chance any cheap device with a ar8316 (i.e. cheaper than a rs pro)? I want to resume working on the driver, but I lack a dev device, and ~100 euros is a bit out of my current budget Oct 30 22:24:02 nope Oct 30 22:26:48 sorry, but can you tell me, what EBU is? Oct 30 22:26:58 external bus unit Oct 30 22:38:06 KanjiMonster: RSPRO 89 USD... where do u live, should be 50 EU Oct 30 22:40:19 florian * r23726 /trunk/package/ep80579-drivers/Makefile: [package] ep80579 depend on their corresponding subtarget Oct 30 22:40:20 OutBackDingo: germany - don't forget VAT, EU is generally more expensive, and I need an additional psu or a poe injector Oct 30 22:40:23 florian * r23727 /trunk/target/linux/x86/net5501/config-default: [x86] resync config-default, enable gpio support Oct 30 22:40:27 florian * r23728 /trunk/package/kernel/modules/other.mk: [package] prefer the cs5535 gpiolib driver after 2.6.32 Oct 30 22:40:32 florian * r23729 /trunk/ (2 files in 2 dirs): Oct 30 22:40:32 [package] add kmod-leds-net5501 Oct 30 22:40:32 Signed-off-by: Philip Prindeville Oct 30 22:40:55 KanjiMonster: can someone donate one by shipping it to you as a gift without it costing you ? Oct 30 22:46:49 OutBackDingo: Never thought about that. I wouldn't decline it, but also I already have a lot on my plate, so I don't know when I would come around to working on it further. Oct 30 22:47:14 A loan device would be fine, too ;) Oct 30 22:47:18 KanjiMonster: well if you want one, i can donate one and have it shipped from Bulgaria Oct 30 23:06:11 [florian]: je viens de t'enoyer un email a propos de net5501 leds. **** ENDING LOGGING AT Sun Oct 31 02:59:58 2010