[02:34] <jburkholder> janimo: poke
[12:42] <janimo> jburkholder, yes?
[13:07] <jburkholder> janimo: hi, I was at the launchpad bugs about sysv-ucontext style functions not being implemented on arm
[13:07] <jburkholder> was looking at
[13:08] <jburkholder> do you know if there's a reason they can't be implemented?
[13:10] <jburkholder> lp #383975
[13:10] <ubot2> Launchpad bug 383975 in debian "[ARM] sysv-style ucontext functions appear not to be implemented on armel" [Unknown,Fix released] https://launchpad.net/bugs/383975
[15:22] <persia> lilstevie: So, firstly, there's a restriction that package names can only be lowercase letters, and a strong interest in having relatively short names.  What do you think about "linux-transformer"?
[15:23] <persia> lilstevie: You'll want to change the Maintainer: I'd suggest "Ubuntu Developers <ubuntu-devel-discuss@lists.ubuntu.com>" unless you have prearranged agreement by the currently listed maintainer.
[15:25] <persia> Something looks to have gone wonky with the header files: these are important to have auto-installed to work around some bugs in DKMS.
[15:27] <emorris> hi, if I want to port some packages for ARM, is it best to download the source from packages.ubuntu.com, or from the SVN or...?
[15:28] <persia> emorris: Most of use use a deb-src entry pointing at onieiric, which ends up being downloeded from ports.ubuntu.com, archive.ubuntu.com, or some mirror.
[15:28] <persia> I'd generally recommend a local mirror, although use rmadison to make sure that you have the current version.
[15:29] <persia> Some folk use the auto-imported bzr trees, which are often correct, although sometimes there are import errors or unuploaded commits, which can lead to code skew.
[15:31] <emorris> persia, cool, I'll probably go for the deb-src way then. Also, what is rmadison?
[15:32] <persia> emorris: It's one of the tools in devscripts: it queries a cgi interface on the archive master to determine which versions of the packages are newest in each release.
[15:32] <persia> Well, each series, but kinda release.
[15:33] <emorris> persia, ah, got it. I tried to Google it, but Google auto-corrected it >:-(
[15:33] <emorris> thanks
[15:33] <persia> Sometimes, for example, packages.ubuntu.com is out of date (by a day or two), because of transient issues, and one can get confused if a package is being rapidly developed and one relies upon that resource.
[15:33] <persia> No problem.  We're always happy to answer questions here.  New folk wanting to help are the ones we like to help the most :)
[15:35] <emorris> Don't know how far I'll get, but it's worth a shot I guess!
[15:35] <persia> What hardware are you using?
[15:36] <emorris> persia, pandaboard
[15:37] <persia> And you're running oneiric?  You probably also want to configure it to use a USB drive for your working area.
[15:37] <persia> (maybe even for everything, depending on how much you like to fiddle)
[15:39] <persia> Did you find the list of packages that aren't building today yet?
[15:41] <emorris> Still on Natty atm. Was only really planning to do a personal project atm, and get packages working as I need them
[15:42] <persia> Aha.
[15:43] <persia> Well, let us know if you find something that doesn't work, especially if you have a fix :)
[15:43] <persia> If you find you have some time, and want to dig more, let us know.
[15:44] <persia> In either case, we're very happy to help get your environment set up for development, and lead you through how we use the packages for development.
[15:50] <emorris> persia, cool, well I'm very happy to help where I can. I haven't really looked into it much yet, but I'll work my way through https://wiki.ubuntu.com/ARM before pestering the channel!
[15:51] <lilstevie> linux transformer is fine
[15:51] <lilstevie> and the maintainer stuff I hadnt changed
[15:52] <lilstevie> re persia
[15:52] <lilstevie> persia: the only reason it is longer is we were working on an <arch>-<platform>-<variant> convention
[16:00] <persia> lilstevie: Right.  So, there's some string limit on these that isn't actually that long.  When one starts to generate all the udebs, this string gets kinda awkward.
[16:00] <persia> As much as I'm a fam of getting as many kernels in the archive as possible, we obviously want to strive for concentration.
[16:00] <lilstevie> heh
[16:01] <persia> It is my expectation that if we can use device-specific kernels to bootstrap specific devices, we'll be able to focus on getting integrated trees for things.
[16:01] <lilstevie> persia: well my only concern is there are 2 variants of the transformer
[16:01] <persia> So, for oneiric, I'm expecting only two tegra kernels: ac100 and transformer.
[16:01] <lilstevie> wifi and 3G
[16:01] <persia> Can a kernel be built that runs on both of them, or do the modules conflict somehow?
[16:01] <lilstevie> I guess a universal could be built
[16:01] <persia> I'd much rather see loadable modules for the networking, and don't care that much about netbooting over 3G, for example.
[16:02] <lilstevie> but given the 3G isn't fully in the wild yet it is hard
[16:03] <persia> emorris: Just let us know if you have questions: we surely haven't gotten all the docs up on the wiki, and we know some bits are confusing.
[16:04] <persia> lilstevie: So, let's pretend we have a combined kernel, and do the wifi module load through udev.  When the 3G becomes more available, we can fix the bugs in that kernel.
[16:05] <lilstevie> persia: well there is a ril driver in all recent versions
[16:05] <persia> Oh, excellent.  Let's pretend that works until we get a bug report.
[16:06] <lilstevie> sounds good
[16:06] <fisuk> argh.. i've generated a rootfs with rootstock, cross-compiled the kernel, but how on earth do i generate the uImage?
[16:07] <persia> lilstevie: I'll be on planes soon for a couple days, but I'll let you know when I am around enough, hoping to build and upload something.
[16:07] <lilstevie> ok :)
[16:07] <persia> THe only other really obvious thing I see from looking over some of the files in debian/ is the changelog: that ought be a real email address, and a real name.
[16:08] <persia> (since you've provided both to LP and in other places, this oughtn't be onerous)
[16:08] <lilstevie> yeah that I can understand :)
[16:09] <lilstevie> I was just getting it to you mainly
[16:11] <persia> That's fine.  I'm happy to have had a look, and glad to report that it looks like you've done it mostly right (from what I can tell from a quick review)
[16:12] <lilstevie> :)
[16:17] <emorris> fisuk, it isn't in arch/arm/boot?
[16:29] <fisuk> emorris: ah, didn't know where to look! thanks!
[16:37] <fisuk> recompiled it just with make... and make uImage did tell the location..  did before with dpkg-buildpackage and somehow missed it.
[16:37] <fisuk> i guess i should write a wiki entry "kernel compilation for idiots"
[18:50] <janimo> jburkholder, hi, from what I read a while agi, ucontext are considered deprecated (not sure by what other API though) and not high priority. But I think just noone got around to implementing it. May require kernel changes too besides glibc, I do not know exactly