**** BEGIN LOGGING AT Sun Dec 31 02:59:57 2006 Dec 31 03:15:10 The way to tell the unslung release is to look in /.unslung Dec 31 03:24:49 Can anyone please point me to how to implement "toolchain" per http://www.nslu2-linux.org/wiki/Debian/FatSlug Dec 31 03:25:42 (Wiki instructions are great but the above was probably written by someone with an obvious software background) Dec 31 04:36:05 03bzhou * r4940 10optware/trunk/make/xrender.mk: xrender: added renderext-stage Dec 31 07:00:30 03bzhou * r4941 10optware/trunk/make/xfixes.mk: xfixes: added fixesext-stage Dec 31 08:28:00 03oleo * r4942 10optware/trunk/Makefile: optware: demode xrender and xfixes for uClibc Dec 31 11:25:52 Anyone for a svn setup question? Dec 31 11:59:27 me Dec 31 12:26:58 03oleo * r4943 10optware/trunk/Makefile: optware: further demote xliraries for uCibc Dec 31 13:51:16 03oleo * r4944 10optware/trunk/make/ (8 files): rebuild libraries for gnuplot Dec 31 15:01:00 03oleo * r4945 10optware/trunk/make/gnuplot.mk: gnuplot: rebuild with fresh libs Dec 31 15:23:34 looks like the ipkg repository for unslung is serving packages for debian -- e.g ipkg.nslu2-linux.org./feeds/unslung/cross/coreutils.6.6-3_armeb.ipk has a coreutils.postinst file that tries to run /opt/bin/update-alternatives. Obviously, this loses on unslung. Any awake who can fix? Dec 31 15:29:41 tlhackque: I thought unslung had update-alternatives? Dec 31 16:25:11 On my unslung slug I get Dec 31 16:25:11 $ type update-alternatives Dec 31 16:25:11 update-alternatives is /usr/sbin/update-alternatives Dec 31 16:25:39 hmm Dec 31 16:25:49 wrong path on one of them Dec 31 16:27:21 well, not on my freshly unslung 6.8. there is no update-alternatives that I can see. But it appears in (at least) coreutils.postinst, cpio, ipkg, and procps... Dec 31 16:28:04 yeah, it should be there Dec 31 16:28:11 especially for coreutils Dec 31 16:28:26 since it replaces several busybox utils Dec 31 16:28:42 Wasn't a problem yesterday - someone do a checkin? Dec 31 16:29:17 no idea Dec 31 16:29:44 03bzhou * r4946 10optware/trunk/ (10 files in 6 dirs): /opt/bin/update-alternatives did not work on nslu2, revert the change Dec 31 16:30:00 aha Dec 31 16:31:29 Indeed. How does the ipkg.nslu2-linux.org server get the corrected files? Dec 31 16:34:55 guys, i was trying to specify the abs path so optware is more self-contained (mainly for slugosbe feed) Dec 31 16:35:30 and it happens that on unslung, ipkg is built-in in the firmware Dec 31 16:35:50 and is *not* self-contained Dec 31 16:36:14 sorry about the problem it causes Dec 31 16:36:39 Bugs happen, but right now the server is handing out uninstallable kits. If someone with sufficient authority is around, can the server be reverted? Dec 31 16:36:58 wait one or two hours, the corrected postinst will be in the feed Dec 31 16:37:50 (I won't lecture about basic testing before release, as I'm sure I'll mess up some year). Dec 31 16:37:50 we're not mission critical, you should not relying on the feed to do mission critical stuff Dec 31 16:38:25 as a workaround, you can copy update-alternatives into /opt/bin Dec 31 16:38:46 OK, thanks. Depends on what you mean by mission critical. I have so many hours of vacation each year... Dec 31 16:39:42 your time is as important as others, i appreciate your patience Dec 31 16:44:33 03bzhou * r4947 10optware/trunk/make/dhcp.mk: dhcp: 3.0.4 -> 3.0.5 Dec 31 16:46:24 And thanks for the dhcp build! Dec 31 16:48:02 i wish we have a process to seperate stable feed from unstable feed Dec 31 16:49:44 something like, 2 or 3 end-user can promote unstable packages into stable feed Dec 31 16:49:57 with vote(s) Dec 31 16:50:37 the centralized model is not going to scale for runtime testing Dec 31 16:54:16 That would be one approach. Another would be for the developer to do basic tests before commit (like, install, run, execute trivial commands). Your idea could work - but some challenges are getting qualified end-users, and timely testing... Dec 31 16:56:23 tlhackque: The "developer testing" idea is the centralized model. There's too many packages and way too few developers to do that. Dec 31 16:56:48 Unfortunately, us devs have lives too ;) Dec 31 17:00:38 Yes, but as a developer - I don't commit anything without at least minimal testing. I agree that we could get more in-depth testing with a decentralized model. But I also believe that part of being a developer is testing what I produce. Bugs will happen anyway - still, one tries... Dec 31 17:01:07 testing is in many cases impossible Dec 31 17:01:21 eg. the zd1211 kernel module for unslung Dec 31 17:01:43 or dhcp (I have a separate dhcp-server, so I can't check without messing up my netwrk) Dec 31 17:02:07 Quite a few packages require specific other hardware or servers Dec 31 17:02:17 which makes it impossible for me to test Dec 31 17:02:32 but I could push a package to unstable for those daring enough to actually test it Dec 31 17:03:04 giving those a simple way to give a "go/no-go" feedback would be enough in most cases Dec 31 17:03:57 I'll just pop in here and emphasize what NAiL said -- I'd dearly love to have all the hardware necessary to test Unslung releases, but the amount of money I can spend out of my pocket is limited. So we must rely on users with the specific hardware to test for the community. It *is* a community effort! Dec 31 17:04:14 It's worth considering. (DCHP - well, I use the failover model between a linux box (secondary) and the NSLU2 primary.) Dec 31 17:04:56 * mwester is still hoping that Santa Claus is just late, and will bring him a 500GB USB drive so we can finally try to track down the Unslung large-drive problem... Dec 31 17:05:39 mwester: Ok, any ideas how to implement this? I don't think it should be much of a problem to set up a "testing" feed Dec 31 17:06:09 It's the feedback and attracting users to this base that is a bit more work. Dec 31 17:06:33 (This can also improve testing for various platforms...) Dec 31 17:06:43 Yes, it's a community effort, and yes, none of us has all the hw. But if one takes responsibility for a package, either you have the hardware, or you have a user who will test... Maybe I'm simple-minded. Dec 31 17:07:52 tlhackque: That would work great. If we were a lot more developers with a lot free time (or was paid for it). I haven't had time for nslu2 stuff for several months Dec 31 17:07:54 (except the odd thing here and there) Dec 31 17:10:43 I have time about twice a year. But when I do have time, I try to do the whole job. If we can make the two-stage release model work, I'd be OK with that too. Dec 31 17:11:44 in the case of coreutils, i did some test, but on slugosbe :-) Dec 31 17:12:02 yeah Dec 31 17:12:30 any1 know how svn tag works? Dec 31 17:12:41 there's way too many platforms to test, and with the coming of the optware-on-slugos thing also another OS to test. Dec 31 17:13:27 what we need, is a web-interface showing the diff between unstable and stable, and allow end-user to vote Dec 31 17:14:09 with sufficient votes, we move the tag from HEAD (unstable) to stable Dec 31 17:14:38 ... for each platform Dec 31 17:14:42 yeah Dec 31 17:14:58 The implementation is the tricky part ;) Dec 31 17:15:43 right Dec 31 17:15:54 You mean, like -- use a version control tool for VERSION CONTROL??!!! :D :D You're right - this is how it should work; i've been surprised how little attention is paid to establishing stable baselines and other configuration management concepts in most opensource development. Dec 31 17:16:52 I'll stop preaching. All I know is that what makes this community work is that things tend to be stable. If we take on more stuff than we (and our helpers) can test, we'll implode. Dec 31 17:16:59 Since I've been considering a binary release of an Alpha unslung (to get the broader testing, as we were just talking about), the idea of "unstable" feeds is really attractive. Dec 31 17:19:49 The missing "ingredient" in this discussion is the infrastructure for the feeds - we need to re-open this topic with the infrastructure members of the core team. (thanks tlhackque for raising the attention level for this.) Dec 31 17:20:19 I was thinking of picking up a 500GB (or maybe a 1TB!) drive for another project., if I can find a decent one at a decent price. I was also thinking about picking up a spare slug. Seems like if I did all that, I might be able to help. What access would you need, and for how long? Dec 31 17:21:23 Actually, all we need is for the owner of said device to equip the slug with a serial port (if their drive exhibits the problem) -- I expect we can get the debugging work done via IRC. Dec 31 17:22:16 (direct access would be nice, but usually impractical, so best left for "if all else fails".) Dec 31 17:24:00 Failing Santa Claus, I'm still hoping to "intercept" a drive for my wife's business, and borrow it for a few days -- I just don't know when (or if) a drive that large might be ordered. Dec 31 17:28:43 I'll have to see what I can scrounge up for serial port components, but I could do that. What failure is being looked at? I know that a long time ago there were tales of data corruption with large disks, but I don't recall a (dare I say) test case... Dec 31 17:31:56 tlhackque: reverted postinst ipk's are in the nslu2 unstable feed Dec 31 17:34:31 thanks! Dec 31 17:35:38 morning [g2] Dec 31 17:36:06 <[g2]> hey NAiL ! Dec 31 17:37:07 [g2]: morning.. happy holidays! Dec 31 17:58:34 03bzhou * r4948 10optware/trunk/make/py-amara.mk: py-amara: 1.1.9 -> 1.2; added py25 subpackage Dec 31 18:09:06 back later Dec 31 18:11:26 mwester:if you want to pursue the disk debug, contact me @yahoo. - Bye Dec 31 20:27:31 * rwhitby just read the feeds discussion Dec 31 20:27:45 the problem is that our "stable" feed on optware is a symlink to the unstable feed. Dec 31 20:27:57 (whereas that is not the case for SlugOS) Dec 31 20:28:31 where does ipkg fetch from by default? Dec 31 20:29:42 stable, I believe. Dec 31 20:30:01 yep, stable. Dec 31 20:30:18 (at least we got that part right :-) Dec 31 20:30:28 In that case, does it matter at all if we yank the symlink and add a testing-like feed there? Dec 31 20:31:36 The way I see it is that this could make it a lot easier for people to test packages specifically for each platform before approving. Dec 31 20:31:50 (given a good interface to do so, of course) Dec 31 20:34:56 I propose a webinterface that checks the differences between the stable and unstable feeds, and lets users approve working upgrades. This shouldn't be an extremely hard thing to do in eg. php. Dec 31 20:35:01 NAiL: why not just do it the way you do it for SlugOS now? Dec 31 20:36:14 because that requires developers to actually test packages with all hardware and platforms Dec 31 20:36:34 slugos is easier since it runs on fewer platforms and is a "clean" system Dec 31 20:37:00 even so, I'm the only one doing maintenance on slugos "stable". Have you seen any commits from me lately? :-\ Dec 31 20:37:14 I like doing the job Dec 31 20:37:35 but if I don't have the time (or the hardware required to test specific packages), there's no way I can do it. Dec 31 20:41:07 Giving unslung end-users the choice of running a (hopefully very) stable firmware and running a "testing"-kind of firmware would hopefully attract people to both camps. I know I'd run one of each on my slugs if I had more than one. Dec 31 20:41:24 (I've donated one to a friend of me who might join development sooner or later) Dec 31 20:42:31 Requiring a package to be "signed-off" by several users before it appears in the stable feed would (imho) be a great way to let the community do a bit of work that is difficult to do by each developers. Dec 31 20:42:38 s/developers/developer/ Dec 31 20:42:38 NAiL meant: Requiring a package to be "signed-off" by several users before it appears in the stable feed would (imho) be a great way to let the community do a bit of work that is difficult to do by each developer. Dec 31 20:43:57 (seeing as we're not that many developers) Dec 31 20:44:26 * NAiL thinks it's a good idea Dec 31 20:44:31 * rwhitby agrees Dec 31 20:44:42 The only problem I see is the implementation of the "voting/promoting" interface Dec 31 20:45:08 I do know a bit of PHP, and I think I could implement such a feature Dec 31 20:45:50 it won't be aestethicly pleasing, but someone else could do the work on that bit ;) Dec 31 20:46:40 I don't see this feature-thing implemented until the next firmware release. I think. Dec 31 20:47:09 since the current release points to stable, then it can be done with the unstable feed location Dec 31 20:47:29 huh? Dec 31 20:47:41 We don't want all users to run unstable? Dec 31 20:48:03 no, just those who are testing packages Dec 31 20:48:08 yeah Dec 31 20:48:30 a vote gets the package from unstable into stable Dec 31 20:48:35 I think it should be an integrated feature in the next unslung version too... Dec 31 20:48:39 eg. a shellscript Dec 31 20:48:55 to do the vote? Dec 31 20:48:58 to make it even easier to vote for a tested package Dec 31 20:48:59 yes Dec 31 20:49:26 that could easily by an ipkg Dec 31 20:49:32 shouldn't be much harder than a wget http://host/script?package=whatever&status=working Dec 31 20:49:36 no need to wait Dec 31 20:50:15 good point Dec 31 20:51:11 Should such an ipkg change the feed address to "stable" by default (incase user has "fixed" something), and then giving the user instructions on how to participate in this "program"? Dec 31 20:51:23 * NAiL will need to re-read this tomorrow. Dec 31 20:51:26 the feed address *is* stable by default Dec 31 20:51:59 all our released binaries point to stable feeds Dec 31 20:52:22 "by default", yes. I've noticed some users changing the feed address to "unstable" in hopes of getting more up-to-date packages. Dec 31 20:52:22 the symlink from stable to unstable is on the feed machines, not the user's machine. Dec 31 20:52:33 that's their perogative Dec 31 20:52:44 I'm just proposing it as a safeguard Dec 31 20:53:10 nah - we don't mess with user preferences like that Dec 31 20:53:29 ok Dec 31 20:56:22 rwhitby:, nail: have a min to disucss the new slugosbe feed? Dec 31 20:58:30 I want to make the barrier a little higher now :) Dec 31 21:01:24 How? Dec 31 21:02:44 * NAiL notes that the slugosbe feed is quite probably almost identical to the slugos-le feed Dec 31 21:17:30 * NAiL wonders how hard it'd be to get optware to build the same stuff LE Dec 31 21:18:39 chacko: is there any reason as to why you're CTCP version'ing me? Dec 31 21:19:26 nail: does not know what you mean, very new to irc here... Dec 31 21:19:58 your irc client has asked my irc client what version it is Dec 31 21:21:33 nail: nothing that I wanted to do :( I checked how long you have been idle, by doing DisplayInfoUser on your nick. Shall I try again just to confirm that is the cause? Dec 31 21:22:01 sure Dec 31 21:22:17 I'm not attributing it to malice, I just wondered why Dec 31 21:22:19 just did it Dec 31 21:22:24 yeah, that showed up again Dec 31 21:22:34 * NAiL paraphrases an old saying Dec 31 21:22:48 "Do not attribute to malice that which can be attributed to inexperience" ;-) Dec 31 21:23:33 That is a bummer ... all I wish to know is if somebody is active or not... I am using IceChat on WinXP. There is nothing here I see idle time otherwise :( Dec 31 21:24:21 np Dec 31 21:24:46 np ?? could u pse try userinfo check on me? Dec 31 21:25:14 I don't have a scripted version like you have. I could try a couple standard things though. Dec 31 21:25:58 Can't remember anything else than version and time actually Dec 31 21:26:05 Yes, it should up as VERSION, and then TIME !! That is ugly to do that to somebody :( Sorry I did not know Dec 31 21:26:25 np Dec 31 21:26:29 ( no problem) Dec 31 21:27:28 I've been on IRC for more than 10 years. Expecting people to know as much as I do is foolish ;-) Dec 31 21:27:33 Could you do me another favour? I am going to go inactive... I dont know after some time it should mark me inactive, and see what you get? Dec 31 21:28:18 I don't know if there is such a feature on this network Dec 31 21:28:36 I think your client just counts since the last thing I said Dec 31 21:28:56 (which is obviously inaccurate as I can talk on other channels) Dec 31 21:29:07 nail: First time I used irc was in '93 under os/2, when my son went to a boy scout jamboree in Holland, have not done much since then up until about a few days ago :) Dec 31 21:29:30 heh, that's a looong time ago Dec 31 21:29:43 I started using irc about that time I think Dec 31 21:29:53 It is even possible my son is older than you :) Dec 31 21:29:57 OS/2 was fascinating ;) Dec 31 21:30:10 Which year is he born? ;) Dec 31 21:30:20 79 Dec 31 21:30:25 is->was Dec 31 21:30:27 hah :-P Dec 31 21:30:32 I'm '78 Dec 31 21:30:34 ;) Dec 31 21:30:49 ok... am over 60 Dec 31 21:31:14 Oh Dec 31 21:31:20 Impressive :) Dec 31 21:31:21 I mean b4 '45 Dec 31 21:32:26 Did research on one of the first time sharing system at UC Berkeley in '67 :) Dec 31 21:32:29 older than my dad actually Dec 31 21:32:42 nice Dec 31 21:33:21 Back to irc. Do you know if somebody is inactive on a particular channel? Dec 31 21:34:58 How one knows? Dec 31 21:35:36 I don't think there's an easy way to figure that out on this network without using bots or a script that sees how long it was since someone spoke last. Dec 31 21:35:40 Dunno... rwhitby mentioned something like getting a msg automatically, and that I should not change my status to idle manually... Dec 31 21:36:15 "away-messages" are frowned upon Dec 31 21:36:34 setting /away if your client doesn't announce it, is a nice thing to do Dec 31 21:36:51 I just set myself as away. Dec 31 21:37:05 some clients announce this Dec 31 21:37:09 I know that now :( Dec 31 21:37:46 I see some msgs, recently from you starting with an * . How did you send that? Dec 31 21:37:59 Those are known as "actions" Dec 31 21:38:06 try saying "/me something" Dec 31 21:38:15 I'm going off to celebrate new years Dec 31 21:38:26 Where r u? Dec 31 21:38:44 Norway Dec 31 21:38:50 so GMT+1 Dec 31 21:39:20 Have a happy new Year! I am in California, it is 2:00pm only. Dec 31 21:39:25 Later all Dec 31 21:39:28 :) **** ENDING LOGGING AT Mon Jan 01 02:59:59 2007