**** BEGIN LOGGING AT Sat Jul 09 23:59:56 2005 Jul 10 01:09:37 NAiL: pong Jul 10 01:09:47 <[cc]smart> hi Jul 10 01:10:05 <[cc]smart> afaik, you know a bit more than about .bb files and OE Jul 10 01:10:23 <[cc]smart> you have a little time to explain FILES_${PN} += and do_stage a little to my questions = Jul 10 01:10:31 <[cc]smart> s/=/?/ Jul 10 01:11:04 <[cc]smart> s/more than about/more than i about7 Jul 10 01:11:05 me? Jul 10 01:11:09 <[cc]smart> yes Jul 10 01:11:19 oh my... I dont know anything, but ok :-) Jul 10 01:11:42 <[cc]smart> FILES_${PN} += includes all files given into the pacakge, right ? Jul 10 01:12:01 yes, but if you use inherit autotools it seems to override it Jul 10 01:12:17 <[cc]smart> that's stage 2. i'm still at stage 1 :) Jul 10 01:12:46 <[cc]smart> do_stage decides which files should go into staging area, and staging area is the source for FILES_${PN} += correct ? Jul 10 01:14:07 what fo you mean by source for? Jul 10 01:14:18 <[cc]smart> staging area is the work/pkgname/image/ directory correct ? Jul 10 01:14:24 yes Jul 10 01:14:33 think so atleast Jul 10 01:14:46 <[cc]smart> source for should mean, it's the place where the functions evaluatng FILES_${PN} += read the files out of Jul 10 01:15:36 <[cc]smart> the current example i'm on is cyrus-sasl Jul 10 01:16:03 <[cc]smart> this has a special owned /usr/lib/sasl2 directory with libraries Jul 10 01:16:17 <[cc]smart> these libraries appear in the image/ directory Jul 10 01:16:40 <[cc]smart> i can include thes into the pacakge by doing: FILES_${PN} += "${prefix}/lib/sasl2" Jul 10 01:16:44 <[cc]smart> but Jul 10 01:16:51 ok, seems right... Jul 10 01:16:52 but? Jul 10 01:17:04 <[cc]smart> in this directory there are also *.la and *.a files which are not really needed Jul 10 01:17:18 FILES_${PN} += "${prefix}/lib/sasl2/*.so*" Jul 10 01:17:34 <[cc]smart> is this the right way ? Jul 10 01:17:43 <[cc]smart> or isn't it that do_staging should become specialized ? Jul 10 01:17:54 or FILES_${PN} += "${prefix}/lib/sasl2/*.so.*" Jul 10 01:18:13 <[cc]smart> since there i see the availability of a library specific command: oe_libinstall -so -a -C lib libsasl2 ${STAGING_LIBDIR} Jul 10 01:18:14 I think you can leave whatever you want in image, just get the right stuff into install Jul 10 01:18:27 <[cc]smart> yes, i think you're right in that it will work Jul 10 01:18:42 <[cc]smart> question is, is this correct by means of OE ? Jul 10 01:18:51 I have no idea what you are supposed to do, the docs are a bit, well, nonexistant Jul 10 01:18:58 <[cc]smart> EXACTLY Jul 10 01:19:10 I guessed that since it works, its ok Jul 10 01:19:17 and someone have to say otherwise later Jul 10 01:19:18 <[cc]smart> sayin this since romans minted the first coin Jul 10 01:19:38 <[cc]smart> i mean, the lack of docs Jul 10 01:19:45 <[cc]smart> ok Jul 10 01:19:54 <[cc]smart> i'll just go with the FILES variable in the case Jul 10 01:20:16 looks like some other packages are doing it that way Jul 10 01:30:37 03ccsmart 07org.openembedded.nslu2-linux * r39ec5c10... 10/packages/ntp/files/ntp.conf: Adding local HW clock and default security. Jul 10 01:30:39 03ccsmart 07org.openembedded.nslu2-linux * rd1dce877... 10/packages/cyrus-imapd/cyrus-imapd_2.2.12.bb: Removed snmp support. Jul 10 01:30:42 03ccsmart 07org.openembedded.nslu2-linux * r8b65cbd4... 10/packages/openslug-init/ (openslug-init-0.10/turnup openslug-init_0.10.bb): Make rmrecovery permanent active per default. Jul 10 01:35:58 <[cc]smart> cool, dis works :) Jul 10 01:36:26 :) Jul 10 01:44:08 03ccsmart 07org.openembedded.nslu2-linux * rb6beeff0... 10/packages/cyrus-sasl/cyrus-sasl_2.1.19.bb: Adding missing plugin libraries to package. Jul 10 02:36:06 DaKa2: pong? Jul 10 03:23:08 [cc]smart: Didn't I already remove snmp support from cyrus-imapd? Jul 10 04:32:38 <[cc]smart> NAiL: yes, you did. but i did before that in my personal copy, before i had write access, too. long before we discussed it even. Jul 10 04:32:57 <[cc]smart> i didn't undo that change. Jul 10 04:33:45 <[cc]smart> i didn't expect that to screw sth. Jul 10 04:33:51 <[cc]smart> i guess it doesn't Jul 10 04:37:38 <[cc]smart> i already love this turbo mod Jul 10 04:37:47 <[cc]smart> slug on steroids, really. Jul 10 04:42:45 it does help the native compile times :-) Jul 10 04:43:34 <[cc]smart> well, i'm already happy about that: Uncompressing Linux.......................................................... done, booting the kernel. Jul 10 04:43:41 <[cc]smart> going by like a flash Jul 10 04:43:50 03ccsmart 07org.openembedded.nslu2-linux * r6db0efe5... 10/packages/cyrus-sasl/cyrus-sasl_2.1.19.bb: Update staging dir for plugin libs. Jul 10 04:45:33 yeah it's noticeably faster - [cc]smart did you make an entry in the turboslug database? Jul 10 04:46:26 <[cc]smart> no, didn't Jul 10 04:46:39 <[cc]smart> is that another of those yahoo thingies ? Jul 10 04:46:45 <[cc]smart> or is that on the wiki ? Jul 10 04:46:56 yes, oh you are one of those no yahoo people Jul 10 04:47:02 never mind Jul 10 04:47:27 <[cc]smart> no, not yahoo specific. i'm one of those "don't want to register an account and income"everywhere Jul 10 04:48:49 income? Jul 10 04:50:06 <[cc]smart> yes, so they know how many penis enlragement pills they can offer me. just one palletteor more Jul 10 04:50:53 I don't recall having to give my income when I registered for yahoo, and if they ask I just lie anyway Jul 10 04:52:29 <[cc]smart> yes, sure. just wanted to stress the idea. it's along the same lines though. you leave a drop everywhere and nowadays there are enough privacy contracts in between all those companies that they can exchange everything, because they all keep it private Jul 10 04:53:06 <[cc]smart> they even don't give it out for free, you have to pay for the data. Jul 10 04:53:17 <[cc]smart> so much do they care about it. Jul 10 04:54:04 so you don't register anywhere? Jul 10 04:54:10 <[cc]smart> nearly Jul 10 04:54:26 <[cc]smart> there are things that i'm more interested in than i refrain Jul 10 04:54:32 <[cc]smart> like slugbug Jul 10 04:55:12 <[cc]smart> but it makes me feel a little dizzy, that on the bug pages you can see the email address right away Jul 10 04:55:33 <[cc]smart> or one you pull the monotone archive and typ monotone list keys you get it too Jul 10 04:57:31 <[cc]smart> ohoh... feels i would float bottom up in a river.... need to get some freight into the belly... Jul 10 05:05:57 oh no, now you made me hungry.. Jul 10 05:24:18 morning, g2 Jul 10 05:24:31 morning g2 Jul 10 05:38:41 <[cc]smart> the turboslug modders should be made aware though, that the effect of the increased power consuption is a bit understated in the howto, htough. Jul 10 05:39:21 <[cc]smart> it DOES get a good deal warmer in the case. one should drill a few bigger holes into that case Jul 10 05:39:42 <[g2]> NAiL, DaKa2 morning ! Jul 10 05:40:15 [cc]smart: I've only had problems with that when the ambient temp was >30C, and two slugs were stacked on top of each other Jul 10 05:40:31 <[cc]smart> [g2]: testslug has cable soldered in and turbo mod applied. will get usb cable monday/tuesday.. laong hte lines of that Jul 10 05:40:37 [cc]smart, how do you know it gets a lot warmer? Jul 10 05:40:40 [cc]smart: I've had three TurboSlugs running continuously (one of them is serving up DVD authoring content), and none of them are any warmer than before. But it is winter here :-) Jul 10 05:40:52 <[cc]smart> i know it gets warmer, cause i have it in my left hand Jul 10 05:40:54 mine doesn't seem warmer either Jul 10 05:41:08 I am comparing to a non-turbo slug Jul 10 05:41:13 <[cc]smart> me too Jul 10 05:41:21 <[cc]smart> have three slugs Jul 10 05:41:25 <[cc]smart> two turbo now, one standard Jul 10 05:41:50 my turboed doesn't seem warmer either Jul 10 05:41:58 I would be more imclined to believe a thermometer Jul 10 05:42:08 could be variation in chips ... Jul 10 05:42:13 and the readings we have say maybe 2C Jul 10 05:45:04 maybe linksys bought all the chips not passing some hard tests at 266Mhz and underclocked them :-) Jul 10 05:48:12 <[cc]smart> i don't think there is risk, but it does get warmer. it's obvious. Jul 10 05:48:28 <[cc]smart> anyways... off for coffee. Jul 10 07:00:40 sigh. Jul 10 07:00:56 * rwhitby comforts dyoung Jul 10 08:13:09 jbot, seen jbowler? Jul 10 08:13:09 jbowler <~jbowler@jbowler.nslu2-linux> was last seen on IRC in channel #oe, 33d 23h 57m 15s ago, saying: 'I've temporarily DEFAULT_PREFERENCE_nslu2 = -1'ed it on nslu2-linux.bkbits.net'. Jul 10 08:13:35 jbot, seen jbowler-away? Jul 10 08:13:35 jbowler-away <~jbowler@jbowler.nslu2-linux> was last seen on IRC in channel #nslu2-linux, 32d 12h 35m 44s ago, saying: '... copy the flex_2.5.31.bb from 1.12 along with flex-2.5.31/include.patch ;-)'. Jul 10 09:28:50 03jbowler 07org.openembedded.nslu2-linux * r0586ecf0... 10/packages/nfs-utils/ (files/nfsserver nfs-utils_1.0.6.bb): Jul 10 09:28:50 Working, tested, nfs-utils. This now pulls in all the dependencies on Jul 10 09:28:50 ipkg install and has a working startup script. Jul 10 09:42:08 [cc]smart: I think you need to increment the PR on ntp_4.20.bb and openslug-init_0.10.bb Jul 10 09:47:37 <[cc]smart> hmm... prolly on init... how do i tell this to monotone ? why do you think on ntp ? Jul 10 09:48:17 <[cc]smart> how can i tell monotone to then not apply the changes to old version, but make it the new one ? Jul 10 09:48:30 [cc]smart: you changed the package (in both cases), so a different version number is required, otherwise what is ntp-4.2.0-r2? Jul 10 09:49:15 Why do you want to create two separate packages (e.g. for ntp?) Jul 10 09:49:51 <[cc]smart> ? Jul 10 09:50:46 You asked 'how can I tell monotone....[to] make it the new one' Jul 10 09:51:08 The implication is that you want to make a new package for ntp - right? Jul 10 09:51:38 <[cc]smart> yes. if i got you right what i should do is copy ntp_4.2.0.bb to ntp_4.2.0-r1.bb Jul 10 09:51:49 <[cc]smart> i would just "add" this prolly. Jul 10 09:51:52 [cc]smart: no, read the bb file. Jul 10 09:52:50 Or, for that matter, look at tmp/deploy/ipk/ntp*.ipk Jul 10 09:54:23 <[cc]smart> ah, ok. Jul 10 09:55:16 <[cc]smart> but then, Q: since the actual change is in files directory. how would it be possible for anybody to use an old version of that. won't that change automatically affect all versions ? Jul 10 09:56:17 <[cc]smart> and to verify: so what i am expected to do is up the r2 in ntp_4.2.0.bb to r3 and commit that Jul 10 09:56:49 yes Jul 10 09:57:51 that way bitbake knows there are changes and will rebuild it Jul 10 09:58:28 [cc]smart: you changed the package - look at the contents of deploy/ipk/ntp_4.2.0-r2_armeb.ipk - if you change a package in any material way you must change the version number (i.e. the new package must not have the same base file name as the old.) Jul 10 09:59:18 A simple rule (which I guess we could implement) is that if the md5sum of the ipk changes something bad has happened. Jul 10 10:00:05 (But it can change on harmless alternations such as comment changes or shared library revisions.) Jul 10 10:00:11 <[cc]smart> how can one look at the content of ipkg files ? Jul 10 10:00:19 ar(1) Jul 10 10:01:40 sigh Jul 10 10:01:54 maybe one of these years oe will build procps again Jul 10 10:03:57 <[cc]smart> cyrus-sasl doesn't have a PR yet Jul 10 10:04:00 <[cc]smart> add one ? Jul 10 10:04:28 <[cc]smart> on the other hand. the way it was before was unfunctional Jul 10 10:05:40 [cc]smart: everything has a PR, look in openembedded/conf/bitbake.conf Jul 10 10:07:21 <[cc]smart> yes, but i mean defined in the .bb Jul 10 10:08:42 03ccsmart 07org.openembedded.nslu2-linux * rb39c1942... 10/packages/ntp/ntp_4.2.0.bb: Upped rev to r3. Jul 10 10:08:45 03ccsmart 07org.openembedded.nslu2-linux * rcf55a1f4... 10/packages/openslug-init/openslug-init_0.10.bb: Upped rev. to r34. Jul 10 10:08:50 <[cc]smart> so nslu monotone installation autmatically merges everything into main ? Jul 10 10:09:30 what is 'installation' in that context and what is 'main'? Jul 10 10:10:15 The PR is defined in bitbake.conf, therefore it is defined in every .bb (because every .bb inherits bitbake.conf) Jul 10 10:10:18 <[cc]smart> from how i read the monotone description i expected that it gathers the changes i apply would got to sth. like a recorded change operation and you sync that change operation up to the server. but it's not applied. Jul 10 10:10:33 <[cc]smart> it's like offered and sbdy then in a second step decides to apply or not to main. Jul 10 10:10:58 I still don't understand what 'main' is... Jul 10 10:11:21 * NAiL looks at the development as of late, and sees that it is good Jul 10 10:11:24 jbowler :) Jul 10 10:11:34 <[cc]smart> main in the sense of "the nslu OE monotone saved source data (bbfiles etc)" you get as default Jul 10 10:12:25 ? So 'main' == 'manifest' in monotone-speak? Jul 10 10:12:58 <[cc]smart> that's my issue. i'm not safe in monotone speak as of yet. Jul 10 10:13:20 If that's correct changes commited (or, for that matter propagated) to a monotone branch update the manifest (actually create a new manifest) of/for that branch. Jul 10 10:13:41 <[cc]smart> creating a new head Jul 10 10:14:00 Don't do that if at all possible. Jul 10 10:14:18 pull, update, commit, sync Jul 10 10:15:02 <[cc]smart> and this doe not create a new head automatically ? Jul 10 10:15:41 Ah, ok. The head changes. monotone heads (it will have a new revision id) Jul 10 10:15:49 <[cc]smart> i expected creating new heads to be the idea of this Jul 10 10:17:02 That's semantically ambiguous in monotone-speak. Creating new revisions is the idea. Jul 10 10:18:08 <[cc]smart> ok, and then i looked at this as AVAILABLE for others, but not merged. so anybody could LATER decide if he wants my change or not Jul 10 10:18:41 No, it's broken if there are two heads on org.openembedded.nslu2-linux Jul 10 10:18:58 <[cc]smart> so there's a mechanism that automerges ? Jul 10 10:19:04 <[cc]smart> after every sync ? Jul 10 10:19:17 It is, indeed, broken. You need to fix it. Jul 10 10:19:37 <[cc]smart> argh Jul 10 10:19:40 <[cc]smart> how i do this ? Jul 10 10:19:45 <[cc]smart> pull prolly Jul 10 10:19:47 <[cc]smart> merge Jul 10 10:19:51 <[cc]smart> and sync Jul 10 10:19:56 <[cc]smart> correct ? Jul 10 10:20:07 (Like I said, this can happen anyway - there is a race, but push, update, commit, sync minimises the time the problem can occur.) Jul 10 10:20:18 pull, merge, sync Jul 10 10:20:28 jbowler: Quick question... openslug-packages is a dependency of openslug-image, right? Jul 10 10:20:35 NAiL: no Jul 10 10:20:47 It's the other way round Jul 10 10:20:51 yes Jul 10 10:20:56 that was what I meant ;) Jul 10 10:21:10 Just not quite here yet :P Jul 10 10:21:25 If you build openslug-packages you get everything that matters for developers. Jul 10 10:21:43 <[cc]smart> fixed now ? Jul 10 10:21:44 See http://www.nslu2-linux.org/wiki/OpenSlug/WhenAndWhereOpenSlugReleasesAreMade (comments invited!) Jul 10 10:21:53 jbowler: It's great. Jul 10 10:21:59 Read it before dinner Jul 10 10:23:51 It doesn't say much about the feed release process - I suspect it may be necessary to make a branch (nslu2-linux.stable?) Jul 10 10:24:37 <[cc]smart> jbowler: monotone repo fixed now ? Jul 10 10:25:13 [cc]smart: yes, that's it, looks good. Jul 10 10:26:21 <[cc]smart> that last open one was cyrus which i'm working on. the state i found it in is: unusable. should i then up revs too ? Jul 10 10:27:04 If in doubt increment the PR. Jul 10 10:27:09 <[cc]smart> ok Jul 10 10:27:53 If the thing doesn't actually build there is not a lot of point, but once it builds someone might have ipkg installed it, at that point confusion will result if a new version has the same PR. Jul 10 10:29:54 NAiL: so I'm saying the package manager has the tricky job of taking stuff from a building-but-not-necessarily-working openslug-packages and snaphoting into the building-and-working feed(s). Jul 10 10:30:54 jbowler: Like I've done with samba/mysql? Jul 10 10:31:14 (which now finally *works* in the stable feed) Jul 10 10:31:22 I thought you checked the changes in to the monotone nslu2-linux branch? Jul 10 10:31:54 They've been checked in to the 2.0-beta release too, when both proved working Jul 10 10:32:15 <[cc]smart> cyrus-sasl should now be operational Jul 10 10:32:23 Oh right - yes, effectively cherry-picking packages from unstable to 'stable' Jul 10 10:32:36 <[cc]smart> and i'm working on cyrus-imapd Jul 10 10:32:40 jbowler: Yeah, I'm gonna look at nfs-utils next Jul 10 10:33:09 nfs-utils is full working... I'm using it on my primary file server. Jul 10 10:33:31 03ccsmart 07org.openembedded.nslu2-linux * r2900e539... 10/packages/cyrus-sasl/cyrus-sasl_2.1.19.bb: Upped rev. to r1. Jul 10 10:33:54 in stable? Jul 10 10:33:55 I discovered the missing startup script the hard way (file server crashes, turboslug, install monotone head, no working nfsd...) Jul 10 10:34:25 NAiL: no, monotone nslu2-linux as of an hour ago... Jul 10 10:34:46 <[cc]smart> is there an example file where i can read how to have an ipkg create directories = Jul 10 10:35:19 [cc]smart: if you think cyrus-sasl is working and it definately builds then add it to openembedded/meta/openslug-packages.bb Jul 10 10:35:23 <[cc]smart> well prolly there is, just too lazy searching if sbdy. just knows one Jul 10 10:35:33 [cc]smart: install -d ${D}/path/you/wanna/make in do_install () (I'm not sure it's ${D} though) Jul 10 10:35:37 [cc]smart: it may be necessary to update the symlinks, NAiL can probably help with that. Jul 10 10:36:18 <[cc]smart> well, it builds and seem to work for me. some 2nd verification would be nice. Jul 10 10:36:27 [cc]smart: install -d "${D}${sysconfdir}/init.d" Jul 10 10:36:50 for example.. creates ${sysconfdir}/init.d (In most cases /etc/init.d) Jul 10 10:37:09 We get the verification by adding it to openslug-packages and waiting for someone to whine about it. Jul 10 10:37:33 I've built cyrus.. I added the symlinks a while back IIRC Jul 10 10:37:35 <[cc]smart> so if i want to create a /cyrus folder i'd do install -d ${D}/cyrus" correct ? Jul 10 10:37:40 [cc]smart: correct Jul 10 10:38:02 <[cc]smart> the cyrus-sasl before today was buildig ok, but the result was not functional. Jul 10 10:38:07 aha Jul 10 10:38:08 Yes, but that doesn't look like the right place to create a directory... Jul 10 10:38:41 <[cc]smart> correct. will prolly move it later. issue is /var isn't correct on openslug to store your mail either. Jul 10 10:39:12 /var is fine Jul 10 10:39:21 ... well /var/spool/mail Jul 10 10:39:37 <[cc]smart> tmpfs on /var type tmpfs Jul 10 10:40:02 Eh... I wouldn't try to store you mail anywhere if you haven't run 'turnup [disk|nfs] -i' Jul 10 10:40:30 <[cc]smart> which i did. but that doesn't change /var or should it ? if it should., turnup is bugyy :) Jul 10 10:40:51 turnup does change /var Jul 10 10:41:18 diff /etc/fstab /initrd/etc/fstab Jul 10 10:42:08 <[cc]smart> root@SCF83824:/cyrus# mount -o bind / /mnt/ Jul 10 10:42:09 <[cc]smart> root@SCF83824:/cyrus# ls /mnt/var/ Jul 10 10:42:09 <[cc]smart> root@SCF83824:/cyrus# Jul 10 10:42:44 [cc]smart: my /var is correct Jul 10 10:42:53 <[cc]smart> tried bindmount ? Jul 10 10:43:00 [cc]smart: what does the diff on fstab output? Jul 10 10:43:14 root@gunslinger:/mnt/var# ls Jul 10 10:43:14 backups lib lock mysql run spool tmp Jul 10 10:43:18 cache local log openldap-data shared test Jul 10 10:43:24 yes. This is after bindmount. Jul 10 10:43:36 <[cc]smart> < /dev/sda1 / reiserfs noatime 1 1 Jul 10 10:43:44 <[cc]smart> > rootfs / jffs2 defaults 1 1 Jul 10 10:44:04 <[cc]smart> or im misusing turnup Jul 10 10:44:20 Can you post the /var line from /etc/fstab? Jul 10 10:44:36 <[cc]smart> but then, since it's usb stick, i'm not sure i wanted to make /var on disk either. Jul 10 10:44:42 <[cc]smart> i think var is good as it is Jul 10 10:45:19 <[cc]smart> this is /etc on (now) root, which is usbstick: tmpfs /var tmpfs defaults 0 0 Jul 10 10:46:35 <[cc]smart> i feel switching /var in between tmpfs and real fs wouldn't be a good thing really (for me). you could decide to mount anything anywhere, and often & for many it will be flash. Jul 10 10:46:45 Yes, it doesn't do setup_var for 'memstick' - maybe it should. Jul 10 10:46:51 <[cc]smart> nonono Jul 10 10:46:58 <[cc]smart> not for me :) Jul 10 10:47:09 What is absolutely certain though is that /var must be used for stuff like mail files Jul 10 10:47:19 <[cc]smart> i would rather suggest to add a directory, possibly /data or /export Jul 10 10:47:38 <[cc]smart> jbowler: for me, disagreed Jul 10 10:47:40 No. That breaks existing convention. Jul 10 10:48:05 <[cc]smart> this is emebdded, not bigbox. Jul 10 10:48:07 it should be in var, if you want it somewhere else why not symlink on just your system Jul 10 10:48:08 * kergoth advises obeying the FHS unless there's a real good reason to do otherwise Jul 10 10:48:15 <[cc]smart> there is no convention to have /var on tmpfs either Jul 10 10:48:17 Exactly... Jul 10 10:48:20 <[cc]smart> but it makes lot of sense Jul 10 10:48:40 Having /var on tmpfs breaks a number of things - indeed it breaks all of the 'state' type stuff. Jul 10 10:48:53 but shouldn't var be on disk even with memstick? Jul 10 10:49:04 Daka2: yes, I believe so. Jul 10 10:49:14 <[cc]smart> yes, tmpfs makes some things complicated. but still makes sense. Jul 10 10:49:14 isn't all the ipkg databases and stuff in var? Jul 10 10:49:19 Right. Jul 10 10:49:34 no, the ipkg databases _belong_ in /var, but they're in /usr due to most embedded distros violating the FHS in that way Jul 10 10:49:45 ah.. Jul 10 10:50:03 <[cc]smart> if /var is to go to disk, it needs a more detailed selection on where to switch to tmpfs Jul 10 10:50:31 <[cc]smart> since it's REALLY not good to just update all so often on flash Jul 10 10:50:32 I believe the 'nfs' case is adequate - it puts /var/tmp on tmpfs, nothing else. Jul 10 10:50:48 <[cc]smart> /var/log ? Jul 10 10:50:54 There's an argument for redirecting /var/log into /var/tmp too. Jul 10 10:51:04 Oh, and /var/run Jul 10 10:51:24 <[cc]smart> then sbdy iron that out. meanwhile i will go to sth. like /data for me. Jul 10 10:52:02 <[cc]smart> just tell me if you want that on montone server or not. for now, sasl became operational and it doesn't need such stuff. for imapd, this is different. Jul 10 10:52:03 Just so long as a released package doesn't do that... Jul 10 10:52:13 <[cc]smart> imapd as comes in bb files is non function ready. Jul 10 10:52:36 <[cc]smart> if i decide for me to go to /data, shall i commit or not ? Jul 10 10:52:42 [cc]smart: why not make it /var in the package, and you make a symlink on your system Jul 10 10:53:04 <[cc]smart> because /var is tmpfs. that won't help anybody until /var is disk Jul 10 10:53:21 var is disk for everybody with disk och nfs as root Jul 10 10:53:23 <[cc]smart> and /var should not be disk before the more detailed selection like /var/log /var/tmp is done Jul 10 10:53:32 [cc]smart: commits shouldn't include system specific changes. Jul 10 10:53:41 s/och/and Jul 10 10:54:23 <[cc]smart> DaKa2: that means, /var is disk for the least percentage of users prolly. Jul 10 10:54:46 does so many people use memsticks? Jul 10 10:54:58 <[cc]smart> jbowler: what is system specific change? Jul 10 10:55:59 <[cc]smart> with the slug, i prefer them a lot, since they are low power and easy handleable plus no moving parts. Jul 10 10:56:00 [cc]smart: to be absolutely clear: don't commit anything which deviates from the FHS unless you are absolutely sure the change applies to all systems - disk, nfs, memstick, ffs2 with mounted /usr etc Jul 10 10:56:28 <[cc]smart> i expect the average person who'd like to have a private mailserver will like me prefer a 1G stick over a XXXGB HD Jul 10 10:56:54 Packages should only add files to the following list (this is from memory, there may be some errors); Jul 10 10:56:58 <[cc]smart> jbowler: this would say, i should NOT commit with mail in /var Jul 10 10:57:08 /etc: configuration stuff only, small Jul 10 10:57:17 <[cc]smart> because mail in /var does definitely NOT make sense for everybody Jul 10 10:57:26 /usr: read-only stuff only (can be shared). Jul 10 10:57:33 but really, turnup should probably be fixed and make just /var/log, /var/tmp and /var/run tmpfs, right? Jul 10 10:57:40 /var: read/write stuff. Jul 10 10:57:45 <[cc]smart> DaKa2: possibly yes. Jul 10 10:57:58 DaKa2: yes Jul 10 10:58:09 <[cc]smart> jbowler: you forgot one condition - /var: read/write/loose stuff Jul 10 10:58:25 loose? Jul 10 10:58:32 <[cc]smart> it's RAM Jul 10 10:58:41 no, you should never loose stuff on var, that is wrong Jul 10 10:59:09 normaly all state information should be there, ie. ipkg database Jul 10 10:59:20 and you dont want to loose that Jul 10 10:59:24 <[cc]smart> ok, write a file into it (while tmpfs sits there) and reboot Jul 10 10:59:36 Daka2: correct, it might be possible to change the ipkg stuff too on just NSLU2 systems. Jul 10 10:59:43 <[cc]smart> choose an important file you have only one copy of for improved effect Jul 10 10:59:50 tmpfs is really wrong on /var Jul 10 11:00:02 [cc]smart: so that would be a bug, right? Jul 10 11:00:12 on a general purpose system atleast Jul 10 11:00:43 <[cc]smart> jbowler: dunno, i learned about openslug using tmpfs for /var and i thought, wel not too bad. i wouldn't want to wear out FLASH like usual. Jul 10 11:01:07 <[cc]smart> jbowler: so openslug devs tell me if it's a bug. Jul 10 11:01:09 hmm Jul 10 11:01:24 <[cc]smart> at least i know, i won't save in tmpfs /var Jul 10 11:01:32 <[cc]smart> my mail Jul 10 11:01:45 which is a really Good Idea (tm). Jul 10 11:02:02 <[cc]smart> so my daring plan was to use /data or /export Jul 10 11:02:16 But when you've turnup'ed, var really should be non-tmpfs Jul 10 11:02:27 imho Jul 10 11:02:35 <[cc]smart> i think, it should be the same as non turnupped Jul 10 11:02:45 <[cc]smart> you can have usbstick Jul 10 11:02:55 <[cc]smart> AND you can lower confusion Jul 10 11:03:09 <[cc]smart> so if there should be a change, it should be good for both Jul 10 11:03:21 <[cc]smart> making it more detailed like /var/tmp Jul 10 11:03:25 You can't fit much mail (alas) in a flash file system with 648kbytes available, and you system will never boot again if you fill it. Jul 10 11:03:32 <[cc]smart> and linking /var log into /var/tmp MIGHT be the way Jul 10 11:03:34 <[cc]smart> i dunno Jul 10 11:03:47 <[cc]smart> i use 1G usbstick Jul 10 11:04:00 If you're running without turnup, on an usbstick, I really think you know what you're doing and can fix this for /your/ situation. I really don't think that will be the "regular user" experience Jul 10 11:04:13 <[cc]smart> and i'll kick every butt that saves his filesystem in my mailserver Jul 10 11:04:25 <[cc]smart> i run WITH turnup on usbstick Jul 10 11:04:38 <[cc]smart> and i LIKE /var/log and others not to be on usbstick Jul 10 11:05:01 [cc]smart: you are re-inventing the wheel. Moving 'must-preserve' read/write stuff out of /var to somewhere else is just plain wrong, the answer is to fix the existing FHS layout so that /var behaves as required. Jul 10 11:05:02 <[cc]smart> but i agree that maybe /var should be on memstick Jul 10 11:05:09 then symlink /var/log to somewhere tmpfs Jul 10 11:05:12 <[cc]smart> if those freequent hangers stay in tmpfs Jul 10 11:05:25 as I said, /var/log, /var/tmp and /var/run as tmpfs :-) Jul 10 11:05:31 yeah Jul 10 11:05:34 <[cc]smart> NAiL: sounds ok, but i think /mtdblock should do the same then Jul 10 11:05:55 [cc]smart: Not a bad idea, actually Jul 10 11:06:00 <[cc]smart> and it should in my personal opinion also stay to the scheme when run from NFS or disk Jul 10 11:06:12 uh, no Jul 10 11:06:28 When you run from NFS or disk, /var/log shouldn't just disappear on reboot Jul 10 11:06:29 no, I want my logs on nfs thank you :-) Jul 10 11:07:01 <[cc]smart> i think per default (which is, for typical user) it should not change behaviour. Jul 10 11:07:06 <[cc]smart> changing behaviour is for devs. Jul 10 11:07:27 <[cc]smart> again, personal opinion Jul 10 11:07:36 yes, we can agree on that part :P Jul 10 11:07:37 <[cc]smart> i don't want to convince anybody of these ideas. Jul 10 11:08:07 <[cc]smart> i just want to convince that for as long as /var is in tmpfs, it should be ok to choose /data or /export or whatever Jul 10 11:08:31 It's a good point that when on flash/usbstick, /var/log, tmp, run should be in tmpfs. But otherwise, I'm pretty sure most users/devs will want those directories persistent Jul 10 11:08:38 <[cc]smart> nad i don't wanna fiddle with tmpfs /var since there is a lot of magic in /etc/init.d to make that work well Jul 10 11:08:45 <[cc]smart> and i've got no clue about that Jul 10 11:09:12 It looks like the modifiable directories are /var/log, /var/run and /var/tmp I haven't got anything else with a timestamp more recent than Jul 8. Jul 10 11:09:53 jbowler: exactly. Those are somewhat frequently updated on an active system Jul 10 11:10:07 exactly, so who fixes turnup? Jul 10 11:10:33 [cc]smart: it's not difficult to understand I think. Look at the way turnup nfs handles /var/tmp, I don't think the other directories will be significantly different. You should probably egrep /etc/init.d to be sure but as I remember it is pretty straight-forward. Jul 10 11:10:38 <[cc]smart> so what you are saying now is: i should wait completing cyrus-imapd, cause some of you will modify /var logic and i should wait till this is done Jul 10 11:11:12 [cc]smart: you need to find a developer who is motivated to support a memstick system. Jul 10 11:11:34 Ideally someone who has experience modifying turnup and the init scripts. Jul 10 11:11:49 <[cc]smart> i motivated to support my memstick system, and i'm gonna put the data somewhere and not fiddle with tmpfs /var Jul 10 11:12:12 I'm sure there's someone on the channel, maybe even someone who has already committed changes to turnup and openslug-init Jul 10 11:12:49 But like I said, so far as other systems are concerned, appearance of new directories would be a bug. Jul 10 11:12:58 <[cc]smart> i actually did, but just a tinzy little fix, so that reboots will stick to expected medium Jul 10 11:13:17 <[cc]smart> which means i won't commit. that's fine for me Jul 10 11:13:32 <[cc]smart> but it also means cyrus-imapd is unusable Jul 10 11:13:36 <[cc]smart> for users Jul 10 11:14:25 [cc]smart: for users with disks/nfs, no. Jul 10 11:14:44 <[cc]smart> you surprise me. compile and run it. Jul 10 11:14:54 <[cc]smart> and then tell me how it goes. Jul 10 11:15:02 ah, you have other uncommitted changes? Jul 10 11:15:25 <[cc]smart> what we discussed before is how to make it compilable Jul 10 11:15:31 <[cc]smart> now we discuss how to make it usable Jul 10 11:15:48 <[cc]smart> cyrus-sasl i think went both paths now Jul 10 11:16:13 <[cc]smart> cyrus-imapd just made the first part, so i'm on preparing a "for use" environment and configuration now. Jul 10 11:16:24 <[cc]smart> that will be the changes uncomitted as of yet. Jul 10 11:16:50 <[cc]smart> one of the things you need for imapd is a repository for mail configuration etc. Jul 10 11:17:02 <[cc]smart> this usually goes to /var, but not on openslug the way it is Jul 10 11:17:21 <[cc]smart> the direct way for me is to put it into other directory Jul 10 11:17:29 no, and that should be fixed in OpenSlug in my opinion Jul 10 11:17:39 not in that specific package Jul 10 11:18:04 [cc]smart: you have said this three or more times three different ways and three other people have said this is the wrong approach Jul 10 11:18:15 <[cc]smart> agreed, i think the /var mod makes sense. just not for me doing it. i'm for now busy enough learning and preparing imapd Jul 10 11:18:35 [cc]smart: I'm *very* grateful for your work, mind you :) Jul 10 11:19:11 <[cc]smart> jbowler: wrong focus. it is the wrong way for the average. but for now it would be the right way for me since i've got no clue what was done all over the place for tmpfs. Jul 10 11:19:11 Yes, I'm just trying to get you to do the maintenance for memstick systems ;-) Jul 10 11:19:25 <[cc]smart> and i do know that there are many things that have been tweaked for it. Jul 10 11:19:41 <[cc]smart> i have looked over it for a basic grasp Jul 10 11:20:18 That's fine - you can maintain local modifications (use local.conf - this is how I maintain my own builds, which always install ntp). Jul 10 11:20:24 <[cc]smart> i'm not feeling stressed with your comments, no issues. Jul 10 11:21:01 good Jul 10 11:21:17 The cyrus-imapd build fails in cyrusdb_berkeley.c Jul 10 11:21:37 <[cc]smart> just need to learn how the branching works in monotone, so i don't accidentally mess these changes in. Jul 10 11:22:00 <[cc]smart> jbowler: surprising, since it has db3 dependncy Jul 10 11:22:23 <[cc]smart> it builds straight through for me without any tricks Jul 10 11:22:32 <[cc]smart> from a clear tmp Jul 10 11:22:41 jbowler: It builds on nudi, fails on my box Jul 10 11:22:56 both running debian etch Jul 10 11:23:05 Then it's probably picking up header files off the build system. Jul 10 11:23:40 Why db3 not db(4.3)? Jul 10 11:24:05 [g2]: You were running nfs? Jul 10 11:24:06 <[cc]smart> how to control these header file stories, these are things i don't know how to handle as of yet. Jul 10 11:24:29 <[cc]smart> but that part of the .bb file was already there Jul 10 11:24:33 <[cc]smart> afaik made by kergoth Jul 10 11:24:48 <[g2]> NAiL, yeah I played around with it Jul 10 11:25:01 <[g2]> [cc]smart, what changed with ntp4.2.0 ? Jul 10 11:25:08 [g2]: Have you tested with jbowlers changes? Jul 10 11:25:08 <[cc]smart> ntp.conf Jul 10 11:25:27 <[g2]> ok... what did you change in there (out of curiosity) Jul 10 11:25:32 <[cc]smart> i just didn't know to up rev so i did that later Jul 10 11:25:35 [g]2, NAiL: ntp not nfs Jul 10 11:25:44 <[cc]smart> added local HW clock as low prio reference Jul 10 11:25:50 <[cc]smart> added basic security Jul 10 11:25:55 18:28 < CIA-8> jbowler org.openembedded.nslu2-linux * r0586ecf0... /packages/nfs-utils/ (files/nfsserver nfs-utils_1.0.6.bb): Jul 10 11:26:05 jbowler: that change :) Jul 10 11:26:42 Ah, ok, too may converstations at once :) Jul 10 11:26:48 yes ;) Jul 10 11:27:22 cyrus-sasl: the missing function is txn_checkpoint Jul 10 11:27:24 <[g2]> [cc]smart, cool. I'll have to check it out Jul 10 11:27:34 dyoung: ping? Jul 10 11:28:17 <[cc]smart> cyrus-sasl compiles and runs fine for me, this is a slug: root@SCF83824:/cyrus# sasldblistusers2 Jul 10 11:28:18 <[cc]smart> smart@smartpal.de: userPassword Jul 10 11:28:21 <[g2]> restrict default nomodify nopeer ? Jul 10 11:28:50 <[cc]smart> argh, me silly Jul 10 11:28:55 <[cc]smart> just posted email Jul 10 11:29:22 <[cc]smart> yes the restriciton and the virtual server for H clock Jul 10 11:30:22 any recomendations on a small footprint dns server? Jul 10 11:30:22 <[g2]> [cc]smart, will that block another slug or PC from getting the NTP time from it ? Jul 10 11:31:05 <[cc]smart> no, this will make it so that other ntp servers (or fakes thereof) cannot modify or influence the timekeeping of the ntp server Jul 10 11:31:08 DaKa2: dnsmasq? Jul 10 11:31:18 <[cc]smart> definitely dnsmasq Jul 10 11:31:25 <[cc]smart> great piece of work Jul 10 11:31:34 <[cc]smart> big bunch of functions Jul 10 11:31:38 NAiL: ahh, thought that was just for wforwarding Jul 10 11:31:44 -w Jul 10 11:31:50 nope, used to run that on my wrt54g Jul 10 11:32:00 <[cc]smart> still do it :) Jul 10 11:32:11 <[cc]smart> gs that is Jul 10 11:32:22 "Dnsmasq is a lightweight, easy to configure DNS forwarder and DHCP server", then that description is wrong... Jul 10 11:33:01 <[cc]smart> it serves out what you have in /etc/hosts, too. Jul 10 11:33:11 ah.. but no zone files? Jul 10 11:33:16 <[cc]smart> correct Jul 10 11:33:20 ahh, dammit Jul 10 11:33:35 <[cc]smart> this is part of the lightweight, and it also makes it so simple to configure Jul 10 11:34:16 I need zone-transfers.. so.. guess Dnsmasq is out Jul 10 11:34:29 <[cc]smart> i'm cluerupted Jul 10 11:34:35 * DaKa2 builds bind instead Jul 10 11:34:56 yeah, you'll probably need bind for that Jul 10 11:36:27 [cc]smart: when you build 'cyrus-sasl' from a clean tmp, what do you build first - openslug-packages or openslug-image? Jul 10 11:36:47 <[cc]smart> openslug-image Jul 10 11:36:59 <[cc]smart> need no more Jul 10 11:37:06 <[cc]smart> after that i choose individual build Jul 10 11:37:17 <[cc]smart> bitbake openslug-image Jul 10 11:37:19 NAiL: that's why it doesn't build. Jul 10 11:37:39 <[cc]smart> then in any order: bitbake postfix bitbake ntp bitbake cyrus-imapd Jul 10 11:38:01 jbowler: Uhm? db3 isn't built you mean? I added db3 as a dep in my local repo, and it *still* didn't build. Jul 10 11:38:07 The package is designed to build against 4.3 (-l db-4.3) but it picks up the db3 headers so it gets the wrong txn_checkpoint Jul 10 11:38:27 oh... Jul 10 11:38:41 <[cc]smart> for me it plain works, i'm just surprised what can i say ? Jul 10 11:38:43 I think I can fix it relatively easily - just a matter of the correct -I Jul 10 11:38:45 so it really depends on db4? Jul 10 11:38:56 Sure as hell tries to link with it ;-) Jul 10 11:39:10 [cc]smart: What confuses me is that it built on nudi and not on my box ;) Jul 10 11:39:32 <[cc]smart> again, for me it jus works :) Jul 10 11:40:04 NAiL: it's order dependent, I believe it will build if db4.3 has not been built. Jul 10 11:40:22 ...because then it manages to pick up -ldb3 somehow Jul 10 11:40:45 Here's the problem line: | ccache armeb-linux-gcc -march=armv5te -mtune=xscale -L/home/work-tmp/jbowler/nslu2/openslug/staging/armeb-linux/lib -Wl,-rpath-link,/home/work-tmp/jbowler/nslu2/openslug/staging/armeb-linux/lib -Wl,-O1 -o sievec sievec.o libsieve.a ../lib/libcyrus.a ../lib/libcyrus_min.a libsieve.a -lsasl2 -lssl -lcrypto -lfl -lresolv -ldb-4.3 ../et/libcom_err.a Jul 10 11:41:05 ugh... so the fix would be? Jul 10 11:41:08 I suspect 'configure' chooses to put in the -ldb-4.3 Jul 10 11:41:15 Just add the correct -I Jul 10 11:41:21 ... I'll fix it. Jul 10 11:41:43 thanks :) Jul 10 11:43:44 <[cc]smart> need to feed the monkey... brb Jul 10 11:44:34 <[g2]> that only takes me about 20 seconds Jul 10 11:44:51 <[cc]smart> like dumoin in a chicken ? Jul 10 11:44:57 <[g2]> DOH! he said "feed" Jul 10 11:45:01 <[cc]smart> s/dumoin/dumpin/ Jul 10 11:45:20 <[g2]> :) Jul 10 11:45:21 <[cc]smart> ye, me, UUUHUHHHAHHAHHH :) Jul 10 11:48:34 03nail 07org.openembedded.nslu2-linux * rc58b0633... 10/packages/meta/openslug-packages.bb: Cosmetic change to OPENSLUG_PACKAGES. Makes it easier to read Jul 10 11:57:09 [cc]smart: Anything else needed in oe-symlinks than cyrus-imapd and cyrus-sasl to build cyrus? Jul 10 11:57:36 I seem to remember there weren't any more Jul 10 11:59:12 <[cc]smart> yes, there was one, but iirc the one was already communicated and added Jul 10 11:59:25 ok Jul 10 11:59:36 <[cc]smart> pcre & postfix were too i think Jul 10 12:00:12 NAiL: looks like cyrus-imapd has the strip problem too... Jul 10 12:00:19 <[cc]smart> i will check mutt, too and let you know. it worked earlier and i'll verify if it still does. Jul 10 12:00:22 jbowler: oh, brilliant.. Jul 10 12:00:35 Now, if only I can find where I put the fix... Jul 10 12:01:56 [cc]smart: postfix isn't in there yet Jul 10 12:02:46 but postfix built? Jul 10 12:03:30 <[cc]smart> if you send a mail to ccsmart@smartpal.de you're sending to a slug. this is postfix Jul 10 12:03:36 <[cc]smart> for postfix you need pcre too Jul 10 12:03:53 pcre was added Jul 10 12:04:05 03repvik * r67 10/trunk/openslug/nslu2-linux/packages/postfix: Added postfix Jul 10 12:04:11 now postfix was too ;) Jul 10 12:10:39 cyrus-imapd should build now (and it will always get db 4.3, regardless of what was built before - although that does depend on the order configure does things.) Jul 10 12:11:42 jbowler: ok, so if I pull and do this: bb -cclean cyrus-imapd cyrus-sasl db3 db4 && bb cyrus-imapd cyrus-sasl Jul 10 12:11:46 It will build? Jul 10 12:11:54 <[cc]smart> then i should rebuild prolly, cause my guess is i'm still using db3 Jul 10 12:12:00 Almost... 'db' not 'db4' Jul 10 12:12:16 Looks like cyrus-sasl is still using db3 though. Jul 10 12:12:30 ok, I'm trying now Jul 10 12:12:46 Hum, actually if cyrus-sasl does what cyrus-imapd did it will use db4 if it is there... Jul 10 12:13:16 first thing it builds is db3, atleast Jul 10 12:13:40 03jbowler 07org.openembedded.nslu2-linux * re8103bb5... 10/packages/cyrus-imapd/ (cyrus-imapd-2.2.12/install-sh cyrus-imapd_2.2.12.bb): Jul 10 12:13:41 Build reliably against db 4.3 Jul 10 12:13:41 Fix the install so that the strip works Jul 10 12:16:13 cyrus-sasl packages (files) liks is broken - /usr/lib/sasl2 isn't in the -dev list. Jul 10 12:17:16 cyrus-sasl builds against db-3.2 - presumably it isn't configured to use 4.3 Jul 10 12:17:45 yeah Jul 10 12:17:56 Yes - configure looks for 4.2 but not 4.3 Jul 10 12:18:06 the build order was db3, cyrus-sasl, db4, cyrus-imapd Jul 10 12:18:40 <[cc]smart> -dev list ? Jul 10 12:18:42 The build order is just determined by the DEPENDS lines, but the DEPENDS need to match what the internal configure actually does Jul 10 12:19:25 hmm Jul 10 12:19:26 On #oe the idea was that maybe staging could be done so that a package only saw what was in DEPENDS, but no one has done that yet. Jul 10 12:19:43 <[cc]smart> NAiL: mutt builds and runs. Jul 10 12:21:14 [cc]smart: No changes needed? Jul 10 12:21:21 <[cc]smart> no changes Jul 10 12:21:36 neat Jul 10 12:21:48 add it to openslug-packages then ;) Jul 10 12:22:01 <[cc]smart> ? Jul 10 12:22:27 packages/meta/openslug-packages.bb, under OPENSLUG-PACKAGES Jul 10 12:23:01 There's a list there over the packages built when you bb openslug-packages Jul 10 12:23:31 03repvik * r68 10/trunk/openslug/nslu2-linux/packages/mutt: Added mutt Jul 10 12:23:50 <[cc]smart> should postfix be added there, too ? Jul 10 12:24:03 yeah, add postfix and cyrus Jul 10 12:24:13 both build for me now Jul 10 12:24:47 <[cc]smart> i think cyrus should maybe wait until there is enough configuration files added so that normal user can see aligt... Jul 10 12:25:19 <[cc]smart> but i don't know the exact idea of it Jul 10 12:25:26 <[cc]smart> so tell me. Jul 10 12:25:39 [cc]smart: That's more or less when I move it to the stable feed ;-) Jul 10 12:25:51 <[cc]smart> ok Jul 10 12:25:56 <[cc]smart> so i'll add cyrus Jul 10 12:26:01 thx Jul 10 12:26:10 <[cc]smart> is there like a rev for that,too ? Jul 10 12:26:25 the openslug-packages.bb? No. Jul 10 12:26:47 It doesn't make an ipk Jul 10 12:27:34 [cc]smart: I'll take a look at cyrus, postfix and mutt when they're build and pushed to the unstable feed Jul 10 12:27:35 <[cc]smart> ah, and then ... Q: cyrus-iapd depends on cyrus-sasl so sasl will autmatically be built when imapd is requested. should sasl still be in the file ? Jul 10 12:28:00 no, that's not necessary then Jul 10 12:28:38 <[cc]smart> mutt is a no brainer, postfix takes a bit of configuration work, sasl is easy but imapd in current state is a hell of manual work :) Jul 10 12:29:02 <[cc]smart> so maybe go for mutt & postfix nut don't get masochistic with imapd yet Jul 10 12:29:25 haha, ok. What is needed for imapd to work? Jul 10 12:30:30 <[cc]smart> two configuration files of medium complexity Jul 10 12:30:34 <[cc]smart> init.d file Jul 10 12:30:54 <[cc]smart> medium complexity directory structure and permissioning for storing data and config Jul 10 12:30:57 ok. the config files are not included in the build? Jul 10 12:31:06 <[cc]smart> not that i found them Jul 10 12:31:14 <[cc]smart> i created own ones Jul 10 12:31:19 03bzhou * 10unslung/ (Makefile make/py-soappy.mk): py-soappy ready for testing Jul 10 12:31:26 ouch... permissioning is a bit harder. Does it require adding users? Jul 10 12:31:34 <[cc]smart> one user Jul 10 12:32:38 <[cc]smart> and the directories are the ones we discussed up there Jul 10 12:32:41 postfix required one user too? Jul 10 12:32:58 <[cc]smart> there was mail group already.... let me recheck about user. Jul 10 12:33:40 adding users has to be done by adding them to base-passwd, since adduser is interactive on the slug Jul 10 12:33:56 <[cc]smart> two users, but i don't remember if they were created by package or if i did them manually Jul 10 12:34:03 you did them manually Jul 10 12:34:05 :) Jul 10 12:34:27 <[cc]smart> would prolly have to echo them in then Jul 10 12:35:11 That'd require a bit too much scripting, I think. If the user already has the user, or if the uid is taken, you might mess up things Jul 10 12:36:14 <[cc]smart> so it needs a little script added to openslug ? Jul 10 12:36:50 If you see in packages/base-passwd/files/openslug/root-home.patch, you'll see what has been done so far Jul 10 12:36:59 cyrus-sasl: it looks like the non-use of db-4.3 is dependent on something aclocal does, so it may be system dependent too. I'll look at it later... Jul 10 12:38:25 It's not the best solution, I agree. Jul 10 12:38:32 <[cc]smart> i'm irritated. packages/base-passwd doesn't have a files directory on my copy Jul 10 12:38:49 uhm... I might be off with the path, took it from mem only Jul 10 12:39:16 yes, replace files with base-passwd-3.5.9 Jul 10 12:39:47 openntpd requires a ntpd user, and mysql requires its own user Jul 10 12:40:01 <[cc]smart> so you suggest a static solution ? Jul 10 12:40:15 <[cc]smart> maybe a mixture Jul 10 12:40:35 <[cc]smart> how about having a file like that where all the POSSIBLE system users a re deifned Jul 10 12:41:01 <[cc]smart> bitbake could then go there and fetch entries and fix them into the packages so that they can get blind echoed Jul 10 12:41:51 [cc]smart: There's a few reserved UIDs, like mysql 64001. Look in the README-file in /tmp/work/base-passwd-3.5.9. If it's there, it's safe to echo Jul 10 12:42:10 (Kinda safe. Check if it's there, echo if it's not) Jul 10 12:42:37 those are UIDs >64000, so they are "safe" Jul 10 12:44:34 <[cc]smart> not in, and relatively short list actually Jul 10 12:44:39 yeah Jul 10 12:46:17 ahh, dammit.. bind provides a /usr/bin/nslookup, but so does busybox Jul 10 12:46:45 guess someone installing bind doesn't really want nslookup.. so.. better remove it from the package Jul 10 12:48:24 <[cc]smart> nslookup is outdated anyways isn't it ? i thought bind now comes with other tools, like resolveip ... ? Jul 10 12:48:57 dig Jul 10 12:49:04 dig, yes Jul 10 12:50:20 file skipped, now I just need some default config-files Jul 10 12:50:45 03bzhou * 10unslung/ (make/iperf.mk Makefile): iperf ready for cross testing Jul 10 12:50:49 DaKa2: keep up the good work :) Jul 10 12:51:04 ;) Jul 10 12:51:40 when I have bind finished I can replace my 1ghz mini-itx with a slug Jul 10 12:52:13 Nehemia M10k? Jul 10 12:52:23 yup Jul 10 12:52:39 Got one of them meself. Brilliant piece of hardware if you switch the fan Jul 10 12:52:43 yup Jul 10 12:52:48 even has i2c on a header Jul 10 12:53:04 It does? Haven't noticed that ;) Jul 10 12:53:07 :) Jul 10 12:53:23 I use on my prototype robot :-) Jul 10 12:53:40 uses i2c to talk to a bunch of ciruits Jul 10 12:53:47 neat Jul 10 12:53:52 thinking of replacing that with a slug to Jul 10 12:53:56 <[cc]smart> roboborgs all over the place. irritating world that is. Jul 10 12:54:08 :)) Jul 10 12:54:39 <[cc]smart> but it's not that you have a mechanical hand yet, that types 100 characters a second, no ? Jul 10 12:54:42 <[cc]smart> :) Jul 10 12:54:54 it can hardly move yet.. so.. no Jul 10 12:55:01 <[cc]smart> would need that Jul 10 12:55:21 You need BIND for your robot?! Jul 10 12:55:33 no, for replacing my other itx Jul 10 12:55:38 I have too Jul 10 12:55:38 ohh Jul 10 12:55:41 two* Jul 10 12:55:48 VoodooZ is using a NSLU2 in his robot as well Jul 10 12:55:51 Tiersten: The robot needs to be able to resolve stuff! ;) Jul 10 12:55:58 <[cc]smart> yes, evasive mobile DNS, ducking and covering from DNS floods Jul 10 12:56:03 lol Jul 10 12:57:15 I just have to make a header for the i2c on the slug and try to protect it somehow, dont want to blow the i2c driver if I connect something wrong Jul 10 12:57:45 <[cc]smart> you've all seen terminator on the show, haven't you ? i'll hold you all responsible :) Jul 10 12:57:53 Use a I2C bus extended Jul 10 12:57:55 extender Jul 10 12:58:07 I just ordered a few to test Jul 10 12:58:16 VoodooZ somehow managed to kill one of the GPIOs which control the I2C bus in the NSLU2. He had to replace it with one of the LED GPIOs Jul 10 12:58:29 03jeanfabrice * 10unslung/ (Makefile make/antinat.mk sources/antinat/prerm): Promote antinat for wiley Jul 10 12:58:29 ouch.. Jul 10 12:58:46 He thinks he might have accidently shorted it to 5V Jul 10 12:59:04 A P82B715 should work Jul 10 12:59:11 <[cc]smart> what do they say ? never fiddle with your system powered up... Jul 10 12:59:25 good, ordered a few P82B715 two days ago Jul 10 12:59:33 should have arrived by now Jul 10 12:59:40 [cc]smart: Unless you're really hardcore and have deep pockets :) Jul 10 13:00:00 <[cc]smart> and depending what you're fiddling with, a grounded iron glove Jul 10 13:00:14 iron glove? Jul 10 13:00:27 I do everything with power on Jul 10 13:00:33 usually works Jul 10 13:01:04 :) Jul 10 13:01:56 <[cc]smart> i had a few shakin experiences. you prolly all know commodore 64/128. it was running like 24/7 at the times and i had it built into a deisgned for that (and self constructed ;)) table/board. Jul 10 13:02:08 <[cc]smart> this made it run warm over time so i removed the covers Jul 10 13:02:36 <[cc]smart> i knew these components like me pocket and so switched on/off everything with eyes closed if must be. Jul 10 13:02:39 <[cc]smart> except for one day Jul 10 13:02:59 <[cc]smart> direct grip onto the powercord leads Jul 10 13:03:09 <[cc]smart> nice little holes in the fingertip Jul 10 13:03:27 ? It was exposed?? Jul 10 13:03:33 <[cc]smart> and elvis was lookin like a wimp on stage compared to me Jul 10 13:04:10 <[cc]smart> well, as i said. the days when we were young Jul 10 13:04:19 <[cc]smart> and did everything Jul 10 13:04:48 ow.. I've only managed to touch 230v a few times, and always grounded close to where I touched it Jul 10 13:05:21 <[cc]smart> it was exposed on the inside of the floppy drive Jul 10 13:05:35 <[cc]smart> power cords lead into the drive then connect over to the power switch Jul 10 13:05:48 <[cc]smart> i touched the back of the power switch in the attempt to switch it off Jul 10 13:06:06 <[cc]smart> and shorted the lines like taht Jul 10 13:06:37 <[cc]smart> another event was more tricky Jul 10 13:06:40 <[cc]smart> pc case Jul 10 13:06:51 <[cc]smart> switched off but power chord left in Jul 10 13:07:02 <[cc]smart> not that atx thingies. the ones earlier Jul 10 13:07:15 at... hated those Jul 10 13:07:18 <[cc]smart> they also lead mains to the front at a switch Jul 10 13:07:41 <[cc]smart> these switches were typically protected with those heat/shrink tubes Jul 10 13:07:51 yup, managed to touch that once, just felt a bit, well.. strange Jul 10 13:07:59 <[cc]smart> i had an HD malfunction Jul 10 13:08:03 <[cc]smart> removed it Jul 10 13:08:05 <[cc]smart> checked it Jul 10 13:08:21 <[cc]smart> reinserted and then, pushed that HD onto that front swicth Jul 10 13:08:54 <[cc]smart> those cobbectors that attached the heads to the logic just disappeared Jul 10 13:09:15 :) Jul 10 13:09:21 <[cc]smart> and i jumped off that tower case which i was bending over on the ground like high jumper Jul 10 13:09:51 always fun to play with mains power :) Jul 10 13:10:03 <[cc]smart> yea, what can i say :) Jul 10 13:12:09 <[cc]smart> in any case, lessons learned :) Jul 10 13:12:26 yeah, always work with mains on? Jul 10 13:12:29 :-) Jul 10 13:12:49 <[cc]smart> well, what good would that powertools be otherwise ? :) Jul 10 13:17:32 I'm getting tired, I've typed 'fidd' to many times now, maybe set up an alias to diff? :-) Jul 10 13:18:45 <[cc]smart> whatever suits :) Jul 10 13:19:40 <[cc]smart> better fidd than goof, no ? Jul 10 13:20:03 <[cc]smart> but you're right Jul 10 13:20:07 <[cc]smart> time to go to sleep Jul 10 13:20:10 yup, but I'll get to that point too soon Jul 10 13:20:16 nahh Jul 10 13:20:17 <[cc]smart> :) Jul 10 13:20:23 <[cc]smart> for me it is Jul 10 13:20:28 <[cc]smart> work is awaiting me Jul 10 13:20:33 nite then :-) Jul 10 13:20:43 <[cc]smart> i'd like to let it wait, but boss doesn't like that Jul 10 13:20:59 <[cc]smart> n8 Jul 10 13:21:05 Im on my vacation, three weeks to go Jul 10 13:21:14 nite nite Jul 10 13:21:17 nite Jul 10 13:21:34 <[cc]smart> ahh, i hat those guys atlking about their holidays.... ;) Jul 10 13:22:15 what? :-) I couldnt stay away from work so I sneeked in yesterday an did some work Jul 10 13:23:12 <[cc]smart> holy holiday-san will send you to holodoomsday for that man Jul 10 13:23:29 haha Jul 10 13:24:18 <[cc]smart> sneaking in for work... i'll have to have a specialized psychiatrist ofr that. Jul 10 13:25:08 I know, there is something wrong with me, could be that I have the best job in the world, I dont know Jul 10 13:25:12 <[cc]smart> but prolly they'll refuse me as lost Jul 10 13:25:36 <[cc]smart> look out for the best girlfriend Jul 10 13:25:54 <[cc]smart> that'll help over this Jul 10 13:26:15 yeah, tried that, never seem to last long.. should probably work more on that part of life Jul 10 13:26:29 some other time Jul 10 13:26:40 <[cc]smart> same with me. know what ? overwork is exactly the reason Jul 10 13:27:11 <[cc]smart> like i did. spent all the time with electrons instead of lady Jul 10 13:27:49 <[cc]smart> but i'll have to understand one way or the other, that electrons will kill me Jul 10 13:27:59 <[cc]smart> if i don't take care Jul 10 13:28:41 yup, one should atleast try to stay alive Jul 10 13:28:57 <[cc]smart> and this has two components Jul 10 13:29:20 <[cc]smart> not just heartbeat and peristaltic Jul 10 13:30:17 [cc]smart: Ok, there's a way to add a user on the fly Jul 10 13:30:27 in do_install ().. safely Jul 10 13:30:29 <[cc]smart> the elctrons are back :) Jul 10 13:30:43 03bzhou * 10unslung/ (Makefile make/py-simpy.mk): py-simpy ready for testing Jul 10 13:30:44 <[cc]smart> all ears Jul 10 13:31:02 adduser --disabled-password --home=/homedir --ingroup Jul 10 13:31:34 * NAiL looks at ByronT flexing muscles Jul 10 13:31:53 <[cc]smart> ok, noted Jul 10 13:32:28 <[cc]smart> but seems like the option that was ruled outbefore... wasn't it ? Jul 10 13:33:25 [cc]smart: yes, because apparently nobody knew how to non-interactively add users using TinyLogin adduser Jul 10 13:34:16 <[cc]smart> so this set of options does that... ok , but now i really need to go. Jul 10 13:34:20 <[cc]smart> thx Jul 10 13:35:41 I think I have a complete bind bb, just a few more tests Jul 10 13:39:59 ah dammit.. I need nfsboot from one network, and another for the external IP.. Jul 10 13:40:10 anyone tried 802.1q vlans? Jul 10 14:02:45 guess not... :( Jul 10 14:04:28 and the ethernet driver is probably broken with large MTU.. oh well. when bind works Ill look into it, maybe works out of the box :-) Jul 10 14:39:40 NAiL: you feel like commiting bind? Jul 10 14:39:52 It works? Jul 10 14:39:56 yup Jul 10 14:40:09 Any config stuff required to make it work? Jul 10 14:40:10 and installs all needed config and starts Jul 10 14:40:13 neat Jul 10 14:40:26 inherit update-rc.d is fantastic Jul 10 14:40:49 http://david.thg.se/saker/bind.tar.gz Jul 10 14:40:53 a bb-file and thre patches Jul 10 14:40:55 three* Jul 10 14:42:19 maybe should build it with a cleen tmp to check if I missed some dependency Jul 10 14:43:11 You didn't put in a MAINTAINER field ;) Jul 10 14:43:19 oh, right.. Jul 10 14:43:23 me? Jul 10 14:43:52 well, you're in the AUTHOR field actually. If you wanna be maintainer as well, I can add that Jul 10 14:44:08