**** BEGIN LOGGING AT Tue Sep 01 02:59:58 2015 Sep 01 07:38:46 morning all Sep 01 12:11:42 hi there, I am quite new to yocts stuff Sep 01 12:11:58 I did my custom layer with some recipes inside Sep 01 12:12:20 while building I am having this warning 'Building libpam but 'pam' isn't in DISTRO_FEATURES, PAM won't work correctly' Sep 01 12:12:58 I know I can workaround by adding 'DISTRO_FEATURES_append = " pam"' to local.comf Sep 01 12:13:25 but how to make recipe and layer kind of self-contained Sep 01 12:13:40 so that I will not have to add anything to local.conf Sep 01 12:26:13 mpas, make a your-distro.conf file in the layer Sep 01 12:34:44 Crofton: something like 'meta-yocto/conf/distro/poky.conf' ? Sep 01 12:35:08 mpas: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#creating-your-own-distribution Sep 01 12:36:44 is it possible to have this without creating a distro? Sep 01 12:37:09 let's say I have a layer containing some packages Sep 01 12:37:39 and end user can choose if he wants to add those to his/her layer Sep 01 12:37:51 or actually distribution Sep 01 12:38:07 I've been thinking that I can add everything to my layer Sep 01 12:38:47 and make it as simple as possible to be used by end user customizing his distribution Sep 01 12:39:14 well pure software layers aren't supposed to influence the configuration - that's supposed to be up to the distro / user Sep 01 12:39:55 so how to deal with pure SW layers? Sep 01 12:40:29 how to fix the issue I am facing? Sep 01 12:42:08 I don't think you can "fix" it per se other than to have the user enable pam as needed Sep 01 13:26:50 how to modify the default location where some specific binary is installed? Sep 01 13:27:50 suppose I want the binary to be placed in /var instead of /usr/bin Sep 01 13:28:08 I assume this will be somewhere in do_install Sep 01 13:36:52 yes... most likely it'll be whatever the software you're building is doing when "make install" is run Sep 01 14:59:28 YPTM: Ready-Access Number: 8007302996 Access Code: 2705751 Sep 01 14:59:46 YTPM: Stephen Joined Sep 01 15:00:31 YTMP: Jussi Kukkonen joined Sep 01 15:00:32 YTPM: Beth joined Sep 01 15:00:39 oO Sep 01 15:00:53 YPTM: Belen joined Sep 01 15:00:54 how do these meetings keeping catching me off guard? Sep 01 15:00:59 * joshuagl will not be able to join, again Sep 01 15:01:33 YPTM: Randy (MacLeod) joined. Sep 01 15:01:48 YPTM: LICENSE/LICENSE-SRC Sep 01 15:01:54 Plans for qt5 in oe-core-2.0 ? Sep 01 15:02:42 YPTM: Michael here. Sep 01 15:03:25 YPTM: Brendan joined Sep 01 15:03:52 vmeson: why? Sep 01 15:07:05 JaMa: qt5 has been discussed before as a replacement to oe-core's qt4 but based on what Stephen is saying now, it seems that there will be on qt upgrade for oe-core this fall. Sep 01 15:08:02 JaMa, Sounds like it's because of the 5th anniversary. Sep 01 15:08:58 the oe-core rename from 1.9 to 2.0 is a 5th anniversary marketing thing. Sep 01 15:11:13 yes qt4 should be moved out, but that's not the reason for moving qt5 in Sep 01 15:11:52 OE is all about layers, oe-core-2.0 should use them as well Sep 01 15:12:37 I suspect it's cause people didn't want to debate if "1.10" was allowed or not.. ;) Sep 01 15:12:57 RP, cant count to ten Sep 01 15:13:17 fray: what is the license reporting tool called ? Sep 01 15:13:20 nobody can, it's 2 digits Sep 01 15:13:30 no it's the math zelots that say 1.10 < 1.9 ;) Sep 01 15:13:32 we should draw a layer depends graph, and work out adjustments to core based on that Sep 01 15:14:14 and other layers Sep 01 15:14:20 it would be fairly easy to do that in theory using the layer index data (which has an API) Sep 01 15:14:27 yes, we should have called Fido 1.008 to give us some space for future Sep 01 15:15:10 * Crofton|work needs an intern Sep 01 15:21:49 * kergoth yawns Sep 01 15:28:10 is there a way to set the hostname in local.conf when building yocto? Sep 01 15:31:45 # The hostname can be changed outside of this recipe by using Sep 01 15:31:54 # hostname_pn-base-files = "my-host-name". Sep 01 15:45:33 moin Sep 01 15:47:16 So BB_GENERATE_MIRROR_TARBALLS packages git repos... but does not do the same for svn repos? Sep 01 15:48:21 who uses svn? Sep 01 15:49:00 yeah, you'd wish... Sep 01 15:49:33 the question here is why this inconsistency, makes the flag a bit useless Sep 01 15:50:02 hmm, what does it do for hg? Sep 01 15:50:03 it wouldn't really be a full mirror tarball like git, it'd just be a tar of a checkout. e.g. you couldn't change revisions without contacting the server. but as long as you included the SRCREV in the tarball filename, it'd be useful. not unlike the work-in-progress git shallow mirror tarball support Sep 01 15:50:09 hg is handled just like git Sep 01 15:50:25 ok Sep 01 15:50:25 both are distributed scms Sep 01 15:50:40 i know Sep 01 15:50:55 any other downloaded tarball is also a specific version... Sep 01 15:51:18 that's not actually the case. it does include the srcrev, but it also includes every commit leading up to that point Sep 01 15:51:22 just nothing after it Sep 01 15:51:54 yes the generated tarball for the git repo Sep 01 15:52:34 I'm saying any source that is not from git but downloaded from a tarball URL is equivalent with "a tar of a checkout" Sep 01 15:52:47 correct. and those are handled quite differently Sep 01 15:53:00 so that seems to be OK for those Sep 01 15:53:29 again, i'm not arguing that we shouldn't do this, i'm saying we need to make sure the tarball reflects what it contains. Sep 01 15:53:30 that's all Sep 01 15:53:50 a git clone can be used fro a number of different revisions, and the repo is shared amongst allr ecipes using that uri. that won't be the case here Sep 01 15:53:57 so the revision needs to be captured in the filename Sep 01 15:54:10 Does it exist any example or skeleton of systemd.bbclass Sep 01 15:54:25 aj_c: what do you mean? tons of recipe sin oe-core inherit systemd.bbclass Sep 01 15:54:27 ok, understood Sep 01 15:54:27 yes, I'd like that Sep 01 15:54:28 I don't mind if there's a new tarball (with a new name) created if the content changes Sep 01 15:54:28 go look at some Sep 01 15:54:31 * kergoth nods Sep 01 15:54:37 that sounds like it'd be of use to me Sep 01 15:54:49 perhaps open a bug in the bug tracker as an enhancement request? Sep 01 15:54:59 perhaps Sep 01 15:56:42 there's an example init-based recipe in meta-skeleton, there should absolutely be a systemd one too Sep 01 15:57:36 joshuagl, thanks man Sep 01 15:59:48 So the real question is about mirroring sources. If all sources are mirrored in git there will be multiple recipes that expect tarballs that need changing. Sep 01 16:00:02 If all sources are mirrored as tarballs and some of those are tarred-up-gits then yocto handles it transparently Sep 01 16:01:21 For various reasons, mirroring the git repo clones that yocto puts in DL_DIR is not as desirable as having all of them as tarballs, making things consistent. Sep 01 16:01:28 git repo clones = the bare repos Sep 01 16:01:38 indeed Sep 01 16:02:25 only inconsistency I have now is one pesky svn repo that is not tarred up. Sep 01 16:03:06 trying to get builds without contacting upstream at all? BB_NO_NETWORK? Sep 01 16:03:23 yes, pretty much Sep 01 16:04:29 trying to manage a mirror of sources corresponding to a particular build Sep 01 16:04:42 I'm toying with a few ideas for both tracking the content of that mirror and distributing that content. Sep 01 16:05:11 git comes to mind :) but it does not scale well with large data Sep 01 16:05:41 Another idea is git-annex Sep 01 16:06:15 Managing this, it would be nice if the format of all sources were consistent, I don't care particularly what that format is. Sep 01 16:18:53 yay, more libgphoto2 induced wreckage.... Sep 01 16:18:56 Computing transaction...error: Can't install libgphoto2-camlibs-2.5.8-r0@core2_64: no package provides libjpeg.so.8(LIBJPEG_8.0)(64bit) Sep 01 16:36:04 paulg, tells me someone upreved libjpeg to a non-compatible version Sep 01 16:37:21 every major libjpeg since 6 has broken compatibility, afaik. i think both our jpeg and libjpeg-turbo are on 8 at the moment Sep 01 16:38:58 the autodep is based on the soname stuff.. so either libjpeg went "missing" or more likely it changed version slightly Sep 01 16:47:53 fray, kergoth - will check in a minute ; got sidetracked with some wreckage on linux netdev Sep 01 16:48:32 since you are using RPM, should be easy to diagnose.. Sep 01 16:48:52 look for the libjpeg* package.. run rpm -qp --provides Sep 01 17:15:35 Hi I've been working on eth0 configuration on my yocto machine. I dont get both ipv4 and ipv5 addresses on eth0. I've setup /etc/network/interfaces to bring up both inet and inet6. Am I missing something in yocto linux configuration ? Sep 01 17:15:51 ipv6 address* Sep 01 17:18:29 Our board is imx28 based. Sep 01 17:20:24 which branch/image? Sep 01 17:21:03 smaller images might not pull in all the kernel modules by default Sep 01 17:21:21 can you modprobe ipv6? Sep 01 17:22:01 when i ran modprobe ipv6 it didnt give me an error Sep 01 17:22:52 what are you building from? Sep 01 17:23:12 * nerdboy building imx28evk on master Sep 01 17:23:14 i followed the freescale instructions to set up yocto Sep 01 17:23:27 linux 2.6.35.3 Sep 01 17:23:32 core-image-base Sep 01 17:23:39 yeah, this build is their layout Sep 01 17:24:02 master builds linux-fslc-4.1x Sep 01 17:24:30 maybe try core-image-full-cmdline ? Sep 01 17:24:39 ok i will try that Sep 01 17:25:10 some days i wish we had include_next/require_next Sep 01 17:25:48 * nerdboy would an extra 5 shots of espresso... some dats... Sep 01 17:26:07 *days even <= hard evidence Sep 01 17:27:16 stray-barn-kitty still likes to hide in the yard Sep 01 17:27:20 a lot Sep 01 17:30:21 Any particular kernel modules I would have to pull in to continue using core-image-base ? I'm building core-image-full-cmdline...but we have space restrictions.... Sep 01 17:31:13 Would like to keep to our minimum config if I can Sep 01 17:45:15 how big is too big? Sep 01 17:45:18 fray, kergoth yeah, we changed from 8 to 9 it seems. Sep 01 17:45:20 commit b749251338972e019a108a988457c60f94032b4d Sep 01 17:45:20 Author: Alexander Kanavin Sep 01 17:45:21 Date: Thu Jun 4 13:21:50 2015 +0300 Sep 01 17:45:21 jpeg: update to version 9a Sep 01 17:45:36 it all depends on your image and config tweaks Sep 01 17:45:36 (our sha, not OE one) Sep 01 17:46:18 akail: lemme paste some stuff Sep 01 17:47:16 bingo.. the problem is the upgrade apparently didn't trigger libgphoto to be rebuilt for you.. Sep 01 17:47:19 cuasing the problem Sep 01 17:47:29 ikely something that should be looked into Sep 01 17:49:46 fray, now that I think of it, I've been seeing these _forever_ Sep 01 17:49:50 NOTE: multiple providers are available for jpeg (jpeg, libjpeg-turbo) Sep 01 17:49:50 NOTE: consider defining a PREFERRED_PROVIDER entry to match jpeg Sep 01 17:50:18 never really gave it a 2nd thought until now.... Sep 01 17:50:22 meta-oe adds libjpeg-turbo that PROVIDES jpeg Sep 01 17:51:04 yeah, it does depend on the config tweaks. I'm also running into an openssh error when building core-image-full-cmdline. Sep 01 17:51:15 https://bpaste.net/show/5c4263424878 Sep 01 17:51:21 Computing transaction...error: Can't install packagegroup-core-ssh-openssh-1.0-r1@all: unable to install provider for openssh: error: openssh-6.5p1-r0@armv5te conflicts with dropbear-2014.63-r0@armv5te Sep 01 17:51:32 it's the tools/debug stuff that makes it "big" Sep 01 17:51:59 did you do it like that? ^^ Sep 01 17:52:11 works fine here on master... Sep 01 17:54:54 this is my local.conf https://bpaste.net/show/f1de5bc33e89 Sep 01 17:56:35 we should really think about switching to libjpeg-turbo as default. it's already the default in debian, redhat, chrome, etc, and it has configure options to select what ABI / SONAME it uses, easing hte compatibility problems Sep 01 17:56:54 take dropbear out Sep 01 17:57:31 you get that by default anyway, so remove that and ssh-server-openssh to IMAGE_FEATURES Sep 01 17:57:37 *add Sep 01 17:58:24 * nerdboy hands kergoth an extra cookie Sep 01 17:58:58 that always goes in my config right away Sep 01 17:59:33 PREFERRED_PROVIDER_jpeg = "libjpeg-turbo" plus native Sep 01 18:23:50 nerdboy: core-image-full-cmdline did not give an ipv6 address either Sep 01 18:36:17 This is my post in community.freescale.com https://bpaste.net/show/e4bf3745e64b. No responses yet Sep 01 18:36:19 looks like it does not have any modules in the sdcard image Sep 01 18:36:26 at least mine does not Sep 01 18:36:57 nor firmware either Sep 01 19:04:37 guys, I've been playing a bit with possibility of installing binaries in other location than /usr/bin Sep 01 19:06:24 I took a look at FILES_${PM} variable Sep 01 19:07:05 then some variations like FILES_${PN}-bin to set specificaly bin location Sep 01 19:07:20 but having troubles doing this correctly Sep 01 19:07:37 I end up having binaries under /usr/bin always Sep 01 19:07:54 I am trying to use meta-linaro-toolchain with fido, I can build core-image-minimal no problem but when attempting to build with -c populate_sdk it fails at the packaging step complaining about libstdc++ dependencies. Sep 01 19:08:29 specifically http://pastebin.com/5UMzZk7h Sep 01 19:10:13 should I take a look at do_package stuff and ovveride something there? Sep 01 19:10:36 or there is some other way? Sep 01 19:11:06 you can put files wherever you want. install them to ${D}/foo/ in do_install, then FILES_${PN} += "/foo", for example Sep 01 19:11:09 also can you point me to some example which is having binaries not in /usr/bin on target device Sep 01 19:11:12 trivial Sep 01 19:11:41 if you want to override the install paths of a recipe, rather than install files explicitly to those paths, you'd have to override our target path variables like bindir Sep 01 19:11:47 depends entirely on what you're trying to do Sep 01 19:12:11 neither sounds like a particularly good idea, but it's certainly doable Sep 01 19:12:43 * nerdboy also wondering what the requirement really is... Sep 01 19:13:08 staylor: trying now Sep 01 19:14:11 could be mssing deps maybe Sep 01 19:14:16 nerdboy: thanks, I'm specifically using linaro gcc 4.9 2015.02 Sep 01 19:14:42 nerdboy: I've also tested fido branch as well as the 2015.02 tag of meta-linaro and both have the same error. Sep 01 19:15:09 maybe it's external toolchain bug then Sep 01 19:16:05 * nerdboy hasn't used external toolchain with current stuff Sep 01 19:16:16 just classic... Sep 01 19:17:06 staylor: are you mucking with IMAGE_INSTALL?FEATURES ? Sep 01 19:18:07 looks like i'm getting new +-cross tools built Sep 01 19:18:17 nerdboy: no though I do remove some distro features "x11 wayland directfb" Sep 01 19:18:46 gcc-crosssdk-initial, etc... Sep 01 19:19:42 staylor: you're using external linaro cross-tools but not their SDK? Sep 01 19:20:19 * nerdboy would recommend keep the "set" together Sep 01 19:20:46 use external both or build your own (toolchain/sdk stuff) Sep 01 19:21:28 with classic we had a locally-hacked sdk task i think Sep 01 19:21:45 with TI stuff, etc Sep 01 19:22:04 yeah that's basically my end goal, though I need to build qt5 cross-tools as well so wanted to start by seeing what populate_sdk would generate. Sep 01 19:22:22 what is your actual requirement? Sep 01 19:22:51 I want to use linaro toolchain, and keep it separate from my BSP sysroots so I can distribute the two separately. Sep 01 19:23:02 * nerdboy guessing it's not "Thou shalt use linaro toolchain for 40 days/nights" Sep 01 19:23:26 you can build both with yocto/oe... Sep 01 19:23:54 ie, full sdk package based on your own image and meta-toolchain package Sep 01 19:23:59 you'd be surprised! the linaro thing is a bit twofold, we can point to we use linaro version X and for people who want to build software from windows they can install the same(-ish) toolchain Sep 01 19:24:27 it's the -ish part that gets in the way sometimes... Sep 01 19:24:43 you can still provide the same things afaik Sep 01 19:24:50 yeah I've built the full sdk without using the external toolchain as well, works fine but I'd like to keep the two separate. Sep 01 19:24:58 you can use the meta-mingw layer to create a YP toolchain (SDK) for Windows as well.. Sep 01 19:25:08 if there isn't a specific reason you need Linaro, it might be causing some of your issues.. Sep 01 19:25:08 plus give commit hashes/other for people to build it Sep 01 19:25:41 * nerdboy trying to get at that ^^ Sep 01 19:25:42 could also work, I started with the external linaro toolchain because it "seemed" to be supported ;-) Sep 01 19:25:59 * nerdboy prefers to support himself Sep 01 19:26:10 just not in public... Sep 01 19:27:32 (I usually like to know what changes are in a given external toolchain and either merge them in or ignore them.. but I'm sometimes a bit of a masochist when it comes to toolchains) Sep 01 19:27:33 :) Sep 01 19:27:57 fray: have you used meta-mingw btw? success/issues? Sep 01 19:28:40 I do use it.. but I havn't in master yet.. I last used in with the YP 1.7 and 1.8 releases and did regularly build SDKs for my windows users Sep 01 19:28:55 (I have some local patches which I don't think have made it in.. [I should check]) Sep 01 19:29:29 fray: thanks, 1.8 is my current target though might be 1.9 by the time we actually get around to supporting windows, but that's good to know it's another option. Sep 01 19:30:44 I also have an unrelated question, or maybe just a clarification, I'd like to allow users to add packages to their target and also pull in the -dev files on the SDK side of things. It seems configuring the opkg (or apt-get) in the sysroot would be a good way to handle this. Anyone have experience with this, pitfalls? Sep 01 19:31:05 checked.. the couple of things I was thinking of are either no longer necessary or have been merged Sep 01 19:31:16 basically trying to avoid re-generating a new SDK everytime a minor package is added or updated. Sep 01 19:31:36 staylor, I don't typically do that.. but I know there has been work done in that area Sep 01 19:32:01 I am open to other ideas for solving this problem Sep 01 19:32:20 opkg offline root should work, in theory... Sep 01 19:32:39 There may be others here who have done it.. it's just not a workflow I typically have done Sep 01 19:32:48 right now I re-package the /opt/poky into a .deb and host a repository. it works but it's silly updating a minor package and needing a 200-some MB .deb file to update. Sep 01 19:32:52 it's just an ar blob with tarballs, can make bash script to unpack in the sdk Sep 01 19:33:26 I always distribute the .zip or .sh to my users.. since they all want to install "somewhere else" Sep 01 19:33:31 (usually a home directory) Sep 01 19:33:53 (.zip you can extract anywhere.. but thats a windows thing) Sep 01 19:34:17 (I'm guesisng it'll be a week or so before I get back to working w/ the SDKs again) Sep 01 19:34:53 yeah I imagine windows will be a whole other bag of trouble Sep 01 19:38:32 * nerdboy feeds the pregnant-kitty-bottomless-pit again... Sep 01 19:46:06 pregnant kitties are not fun when they go into labor on memorial day.. and need a C-section... Sep 01 19:46:14 ask me how I know that.. :P Sep 01 19:46:41 (thank god I live in a rural area with a rural vet who doesnt charge a fortune for emergency procedures) Sep 01 19:46:56 (and who knew a cat could need a C-section) :P Sep 01 19:49:43 ... (of course if I didn't live in a rural area, I also probably wouldn't have cats...) Sep 01 19:59:48 Any pointers on the ipv6 imx28 issue ? Sep 01 20:12:41 do you have modules? Sep 01 20:13:07 what are you using for rootfs? tarball? sdcard image? Sep 01 20:13:46 # modprobe ipv6 Sep 01 20:13:46 modprobe: FATAL: Module ipv6 not found. Sep 01 20:15:16 netconfig has ipv6 Sep 01 20:16:32 # zcat /proc/config.gz | grep IPV6 Sep 01 20:16:32 # CONFIG_IPV6 is not set Sep 01 20:16:50 i dont get the fatal error Sep 01 20:16:57 for modprobe Sep 01 20:16:57 ^^ Linux imx28evk 4.1.2-fslc+g95d9e15 Sep 01 20:17:15 check your config Sep 01 20:17:34 and make sure you have /lib/modules/whatever Sep 01 20:18:10 nand ubifs image for rootfs Sep 01 20:18:29 i dont have /proc/config.gz Sep 01 20:18:48 modprobe config(s) ? Sep 01 20:20:33 i have /lib/modules/2.6.35.3-maintain+yocto+gb3912bb Sep 01 20:22:50 no modprobe.conf Sep 01 20:24:44 linux-imx/2.6.35.3-r45/.git/.config has CONFIG_IPV6=y Sep 01 20:25:41 then you're missing config and/or tools? Sep 01 20:26:05 that's the part i dont know...:-( Sep 01 20:26:10 * nerdboy has to go back and modify kernel config Sep 01 20:45:31 fray: she adopted us last week Sep 01 20:46:04 so, homeless-pregnant-kitty-eats-like-a-german-shepard Sep 01 20:46:42 nerdboy, move to a rural area.. you get to experience that about once every 6 months or so.. :P Sep 01 20:46:51 spring and fall.. kitten everywhere.. Sep 01 20:46:57 barn kitties... Sep 01 20:47:07 yup.. Sep 01 20:47:28 since living here for 8 years we've dealt with about 8 litters... Sep 01 20:47:55 just had one this spring.. the cats/kittens all disappeared last week and we havent seen them like we were.. (but they're still eating the food we put out for them at night) Sep 01 20:48:09 so I think mom-cat has probably nearly finished night hunting training Sep 01 20:51:08 kinda surprised how spry she is at that size... Sep 01 20:51:46 first one for her, she's not quite a year old give or take Sep 01 20:51:49 fray, how many litters per gallon again? I'm metric. Sep 01 20:52:32 nerdboy thats the one that needed the C-section for us.. :P Sep 01 20:53:09 she was just about a year old when she had her litter.. water broke on my lap.. and she spent two hours in labor, not kittens... thats when we got to go to vet on memorial day.. :P Sep 01 20:53:24 yay Sep 01 20:53:29 (we got her fixed as soon as the kitten were weened) Sep 01 20:53:42 ($250 for 4 kittens.. not bad for an emergency visit) Sep 01 20:54:04 would have been $50 more, but my other half assisted in the surgery... Sep 01 20:54:11 if cats (or humans) really had that much trouble there would be a lot less of them Sep 01 20:54:54 vet said this one was in trouble for sure... her labor had stopped by the time we got to the vet.. kittens (and her) survived only because of the c-section Sep 01 20:55:15 the vet didn't say we had to come back Sep 01 20:55:35 ya.. normally no reason for a vet to be involved in kittens.. :) this was 'unique' Sep 01 20:55:36 just "here's the bill" and "good luck!" Sep 01 20:56:22 (luckily we found homes for all of the kittens in that litter.. I know 2 of them are doing well.. no idea about the other 2...) Sep 01 20:56:44 and the mom-cat is still with us.. (she's the one that follows me around the house meowing..) Sep 01 20:56:56 ...and one of my dogs likes licking... :P Sep 01 20:56:58 i suppose that must be easier than "helping" am obstructed elephant poop... Sep 01 20:57:24 (it's bad when the dog gets a hairball from licking the cat(s)) Sep 01 20:57:52 * nerdboy probably watches too much Nature on PBS Sep 01 21:17:05 is there something similar to bitbake world but that ignores packages missing dependencies (because they're disabled or removed, such as x11). Sep 01 21:25:36 So I think I found the mysterious bug in pseudo 1.7.0. Sep 01 21:26:00 When I changed path canonicalization not to do allocs, I removed a lot of free()s that were no longer relevant. Sep 01 21:26:12 I did not think of the case realpath(path, NULL) where I was not freeing the thing because I was *relying* on it being a new allocation. Sep 01 21:27:00 Still testing more things, but that should fix at least the obvious ones that were blowing up deep in rpmdb around Realpath(). Sep 01 21:38:49 3 **** ENDING LOGGING AT Wed Sep 02 02:59:58 2015