**** BEGIN LOGGING AT Tue Dec 03 02:59:58 2013 Dec 03 08:04:52 good morning Dec 03 09:48:55 morning all Dec 03 10:00:56 Crofton|work: if the linux-dummy patch works for you, please add your acked-by to it Dec 03 10:02:01 bluelightning: nice work on the buildhistory/pkgdate patchset! Dec 03 10:02:17 koen: thanks! Dec 03 10:02:33 I meant to do some benchmarking before sending actually Dec 03 10:59:07 morning all Dec 03 11:03:45 morning pb_ Dec 03 11:08:35 hi bluelightning Dec 03 11:08:57 does anyone know why my piglit/python-numpy patches weren't merged into meta-oe? Dec 03 11:31:27 koen, ok Dec 03 11:31:58 since I blindly tested, should I add tested-but? Dec 03 11:33:19 tested butt? Dec 03 11:52:39 obviouslt the recent shopping surge had me thinking buy Dec 03 11:56:13 koen: i don't suppose you recall why you added the generic blit patch in http://cgit.openembedded.org/openembedded/commit/?id=a0c6d9cc9bb2f9691cbedbd263c92e00106562c6? :) Dec 03 11:58:07 let me see Dec 03 11:59:18 rburton: iirc it sped up scrolling a lot with the matching xorg patch Dec 03 11:59:50 that's pretty good recall :) Dec 03 12:00:25 considering it was four years ago Dec 03 12:01:59 http://cgit.openembedded.org/openembedded/commit/recipes/xorg-xserver?id=bdc1442952b8ab1bec976d6aae7afc5a36365495 Dec 03 12:02:05 that's the matching Xorg thing Dec 03 12:02:34 the xorg and pixman guys haven't accepted this, not sure what the 'correct' way is nowadays Dec 03 12:02:46 maybe ssvb knows Dec 03 12:04:01 the line from pixman upstream is that pixman is for fast-paths only, a generic blit is assumed to be done in the calling code Dec 03 12:04:40 pinged the list last night with the pixman portion, it was rejected. Dec 03 12:05:06 koen: top marks for recollection though! Dec 03 12:05:41 well, as we don't have the matching xorg patch in oe-core anyway, incredibly tempted to drop it. Dec 03 12:05:48 it was something I spent a lot of time on :) Dec 03 12:06:01 yes, JaMa and I decided not to support it in xorg anymore Dec 03 12:06:40 it's sad, but upstream doesn't want it and noone has the time to support it Dec 03 12:07:05 it is a massive speedup on 'dumb' x drivers Dec 03 12:07:56 rburton: ask your coworkers about reverse-blits on atom, that's a fun story about test methodology Dec 03 12:09:20 yeah that's familiar but i can't recall any details Dec 03 12:09:32 clearly i don't have koen-level memory Dec 03 12:11:47 it turned out to be a dud, but it did spawn a lot of discussion and testing on the pixman lists Dec 03 12:13:28 rburton: scrolling performance was a big issue for a customer and later TI decided to buy a closed source blob to help with that Dec 03 12:13:40 so I have a vivid recollection of scrolling related things Dec 03 12:13:57 but ask me what I had for dinner yesterday and I'll have to think hard Dec 03 12:14:10 i recall nokia being obsessed with scrolling Dec 03 12:14:40 yes Dec 03 12:14:58 nokia needed kernel patches to allow floating point stuff in signal handlers Dec 03 12:15:09 otherwise you'd get massive corruption during scrolling Dec 03 12:15:56 scrolling is one of the things noone thinks about till it's too late Dec 03 12:16:50 just look at how many rewrites android had for that issue and it still stutters Dec 03 12:17:53 rburton: btw, I did it a bit differently when chopping down the buildhistory repo Dec 03 12:18:22 rburton: after letting it run for 4 days I aborted it and moved it to an ssd, that only took 1.5 days Dec 03 12:18:54 ! Dec 03 12:19:00 the git gc should help a lot Dec 03 12:19:22 it shrunk my 3G repo down to 500M and reduced git status time from cold from 2.5 minutes to 0.5 minutes Dec 03 12:19:41 http://pastebin.com/p6DHqHpB Dec 03 12:19:56 that and git gc helped a lot Dec 03 12:20:18 the above is analogous to git graft, but more error prone Dec 03 12:20:24 heh Dec 03 12:21:00 i did wonder if there's a way of injecting the empty tree sha into a rebase to throw away the prior history Dec 03 12:38:33 Anyone can comment about what's happening with the patchset adding bash to many recipes? Nobody ever noticed that? Dec 03 12:38:55 "Fix the problem of implicit bash dependency" Dec 03 12:39:48 * ant_work fears to find bash in the cpio in the next days... Dec 03 12:47:52 ant_work: similar patches have been showing up for perl as well Dec 03 12:48:10 ant_work: btw, any plans to integrate the armv8 patches for klibc? Dec 03 12:49:07 last I tried, it was really too early for arm64 Dec 03 12:49:44 there are pending patches in the ML Dec 03 12:50:05 arm64: Add arm64 support Dec 03 12:50:05 Nov 12 Dec 03 12:52:48 https://bugs.launchpad.net/linaro-oe/+bug/1081162 also has an overview Dec 03 12:53:26 you can try master http://git.kernel.org/cgit/libs/klibc/klibc.git Dec 03 12:53:36 if it works we can backport the patches in klibc Dec 03 12:59:49 koen: do you test mipsel by chance? Dec 03 13:05:23 nope Dec 03 13:06:09 seems also fixed Dec 03 13:24:16 ant_work: they won't be blindly merged, fear not :) Dec 03 13:26:34 * koen tries to work up motivation to get back to perl/cpan recipe debugging Dec 03 13:30:19 that reminds me Dec 03 13:30:30 Stygia: hi, any progress on the meta-perl stuff? Dec 03 13:45:31 Stygia: there is still quite a few patches on patchwork for meta-perl Dec 03 13:47:10 bluelightning, Hey. Dec 03 13:47:16 Stygia: hi Dec 03 13:47:25 bluelightning, Sorry, but no. I should just stop promising you things, I've unfortunately been really busy lately. Dec 03 13:47:59 Stygia: ok Dec 03 13:48:02 I will try and set aside time on weekends, but we have a launch coming up that can make or break us. Dec 03 13:48:21 [koen@rrmbp meta-perl]$ git cherry -v | grep perl-module | wc -l Dec 03 13:48:21 17 Dec 03 13:48:43 I'll wait a bit before sending those into patchwork Dec 03 13:48:48 * koen stabs cpan*.bbclass Dec 03 13:49:00 Hmm. Dec 03 13:49:04 I will try and get it done, actually. Dec 03 13:49:26 I've been saying I would for too long, I'm sorta embarrassed at this point. And I have been 'hoarding' quite. a few perl recipes, so. Dec 03 13:50:18 WIP is here for everyone else's reference: https://github.com/EmilRP/public_bb_recipes Dec 03 13:51:14 why is everying lib${module}-perl ? Dec 03 13:51:21 I sent a pull request with some cleanups; the plan was to get the recipes into shape and then send patches to add the cleaned recipes to meta-perl Dec 03 13:51:34 I should just look at this now actually. Dec 03 13:51:34 koen: because that was the convention established in OE-Classic years ago... Dec 03 13:51:48 drat, need to start renaming my recipes Dec 03 13:52:10 Well, once I fix this thing so /var/log is non-volatile in my image, I established it was initscripts messing with that and not just base-files Dec 03 13:52:38 I think the naming convention was derived from Debian Dec 03 13:52:45 Yes, that is very debian-esque Dec 03 13:56:21 I don't recall what was left to do with those recipes actually Dec 03 13:56:29 it could be that some of them still don't build properly Dec 03 13:57:28 Some of them are a bit messy. Dec 03 14:00:05 building properly is only one side of it Dec 03 14:00:27 koen: well, it's a first step, if it doesn't build there's no point worrying about anything else ;) Dec 03 14:00:36 right :) Dec 03 14:01:01 Well, most of them do, though. Dec 03 14:01:04 I am actively using these recipes. Dec 03 14:01:08 Most of them anyway, but yea. Dec 03 14:01:12 They are hacks in some places. Dec 03 14:01:33 I'll try and look at it this week. Dec 03 14:01:43 Actually, let me just tell my GF to remind me, that's how I get shit done. Dec 03 14:03:24 Alright. I'll still do it gradually, and it may not be before January. But I'll be sure to gradually set time aside for it. Dec 03 14:03:39 maybe the answer is not to try to do them all at once Dec 03 14:03:49 Yes, that's what I'll do, do them gradually. Dec 03 14:04:30 I suspect there are a number in there that are fine at the moment (at least after my cleanups) Dec 03 14:07:26 Yes, there really should be. Dec 03 14:07:30 At least the more recent ones I've made. Dec 03 14:07:52 Indentation and such aside, though. Dec 03 14:10:44 bluelightning, I will try and get it done, I'm sorry about sucking so much at delivering. Dec 03 15:57:51 any cmake "experts"/users here? Dec 03 15:57:57 got a patch to cmake.bbclass that needs review Dec 03 15:58:22 rburton: does it print "use a real buildsystem instead!" and die? Dec 03 15:58:27 sadly not Dec 03 15:58:28 if not, you should add that Dec 03 16:04:45 YPTM: Paul Eggleton is on Dec 03 16:05:06 YPTM: Beth Flanagas is on Dec 03 16:05:15 Flanagan, that is Dec 03 16:05:19 bluelightning_: wrong room Dec 03 16:05:24 argh Dec 03 16:05:55 could be worse, could be a password Dec 03 16:06:06 heh Dec 03 18:18:20 Does OE complain if I in my recipe apply a patch that already has been applied in the upstream source ? Dec 03 18:19:02 it should fail to apply Dec 03 18:19:04 it just runs patch. patch will fail Dec 03 18:19:28 it will also likely report that the patch can be reverse applied, indicating that it's already incorporated Dec 03 18:26:39 ok, so no need for me to regularly check wether patches have been upstreamed or not then Dec 03 18:26:43 thanks Dec 03 19:55:12 What can I use to generate an encrypted password (for /etc/shadow) so that busybox's login understands? I want to set a default password, and I intend to do that via usermod (ROOTFS_POSTPROCESS_COMMAND). Dec 03 20:29:41 Does OE ship with a command for removing "old" images in all the build/tmp-eglibc/deploy/images/ subdirectories ? Dec 03 20:32:10 jkroon_: nope, best to use rm if you need to do that Dec 03 20:32:52 kergoth, ok Dec 03 20:37:44 Nevermind my latest question. I could make it work. **** ENDING LOGGING AT Wed Dec 04 02:59:58 2013