**** BEGIN LOGGING AT Fri Jun 03 23:59:56 2005 Jun 04 01:51:39 morning all Jun 04 01:51:59 good morning RP Jun 04 01:52:43 I guess I should pay some attention to the c7x0 this weekend :) Jun 04 01:53:08 has it gathered a huge pile of dust? Jun 04 01:53:39 I'm using it for reading books but the kernel work has slipped back a bit :) Jun 04 01:53:53 reading?!?! Jun 04 01:54:09 The 770 isn't going to do much until I can get working wifi on it... Jun 04 01:54:16 * koen has books for that Jun 04 01:54:48 books are far too last century ;-) Jun 04 01:55:02 (paper ones that is) Jun 04 01:55:20 RP: you could install your own image on the N Jun 04 01:56:08 RP: http://handhelds.org/~gpe/gallery/view_photo.php?set_albumName=Various-Devices&id=port_24090 Jun 04 01:56:18 that's a complete OE image Jun 04 01:56:54 koen: nice work. How much of the device's functionality is supported? Jun 04 01:56:57 * ljp needs a 770 Jun 04 01:57:12 it uses the stock kernel + initrd Jun 04 01:57:24 so the hardware should work Jun 04 01:57:29 working wifi? :) Jun 04 01:58:02 I think florian used wifi to scap it, but I'm not 100% sure Jun 04 01:58:27 RP: could you extract a dmesg and/or bootlog from the 770 and mail it to me? Jun 04 01:58:54 koen: No, because I have no way of connecting to the thing :-( Jun 04 01:59:03 not even usbnet? Jun 04 01:59:09 Tat's why I need wifi so much... Jun 04 01:59:21 It didn't seem to want to do usbnet last time I tried Jun 04 01:59:26 I'll give that a go again though Jun 04 02:00:09 I think it just shares the contents of the rsmmc card as standard Jun 04 02:01:39 does anybody know a tool to view .sxi slides in osX? Jun 04 02:02:04 and no, OOo doesn't work on osX Jun 04 02:46:44 RP: is http://handhelds.org/moin/moin.cgi/HpIpaqHx2000 up to date? Jun 04 03:00:46 good morning pb_ Jun 04 03:12:09 koen: Mostly. I need to update some of the gpios Jun 04 03:13:22 koen: Unless there's something I've missed? Jun 04 03:20:55 koen: good morning Jun 04 03:23:18 RP: the wiki mentiones something about "complete unknown" and cf Jun 04 03:24:17 koen: I suppose I could put my guesses down :) Jun 04 03:30:03 is there some way to see what bitbake is trying to do in the "do_fetch" stage? Apparently fetching of "quilt-native-037" succeeds, but in fact it doesn't fetch anything Jun 04 03:31:15 mardy did you look at the stamps-directory maybe fetch ist already done Jun 04 03:33:51 woglinde: it's strange, there is no "stamps" directory: just "cache" and "work" Jun 04 03:34:12 mardy strange Jun 04 03:36:53 Err, how do I specify a defconfig image for the kernel ? Jun 04 03:41:08 aboeglin: I think it's specified in the machines conf files Jun 04 03:41:43 aboeglin: but I shouldn't need to change them, just set the right value for the MACHINE var in your local.conf Jun 04 03:43:57 Actually, I'm trying to build handhelds-pxa-2.6_cvs for hx4700 ... I guess I have to modify the do_configure a little Jun 04 03:46:10 aboeglin: check packages/linux/handhelds-pxa-2.6/* Jun 04 03:46:26 koen: I've updated bits of it including the gpios. Let me know if I've missed anything Jun 04 03:47:33 RP: thanks Jun 04 03:49:38 koen: I've been taken off the hx2750 for a while... Jun 04 04:20:13 hi Jun 04 04:26:47 hi CoreDump|home Jun 04 04:29:16 afternoon Jun 04 04:29:19 reenoo_: I'm trying to add rxtx to OE, any advice on that? It fails on a missing classpath Jun 04 04:31:47 umm.. what's rxtx? Jun 04 04:32:15 http://rxtx.org/ Jun 04 04:32:28 a serial stack for java using native stuff Jun 04 04:34:32 reenoo_: is there a non-hackish way of exposing a classpath to an OE build? Jun 04 04:34:32 ah, right. I guess you could start with staging gnu classpath Jun 04 04:56:15 Err, how long does it takes to build a gpe-image ? (I'm on a 1.3 GHz Celeron M) Jun 04 04:56:26 hm 3 hours Jun 04 04:56:36 maybee 4 Jun 04 04:56:45 ok, thanks Jun 04 04:57:52 aboeglin: from scratch? Jun 04 05:01:01 yes Jun 04 05:01:19 I just passed the kernel step Jun 04 05:01:33 ah,3 or 4 hours after that Jun 04 05:01:41 depending on the speed of your disc Jun 04 05:02:22 Ah ... I'm using a laptop, so it will be a little more, then. Jun 04 05:03:36 *yawn* morning guys! Jun 04 05:11:17 * reenoo_ eyes at CIA-4 Jun 04 05:11:36 gnutls: export asn1_* functions from libgnutls.so. fixes bug #57. Jun 04 05:17:42 sylpheed: actually enable gnutls. remove gpgme from DEPENDS for now since it's not enabled and doesn't work either ATM. Jun 04 05:18:04 CIA-4: ping Jun 04 05:21:43 any news on the SCM front? Jun 04 05:27:44 reenoo_: be careful, sylpheed's configure script needs some m4 macros from gpgme, irrespective of whether it's configured in or not. Jun 04 05:27:56 if you want to build without gpgme on hand at all, you will have to patch the configure.ac to remove those references. Jun 04 05:28:26 oh Jun 04 05:28:29 * reenoo_ fixes Jun 04 05:29:31 * pb_ heads to the tool hire shop Jun 04 05:29:32 later all Jun 04 05:29:39 later pb_ Jun 04 05:30:05 ~users Jun 04 05:30:07 so this is what we get for sharing our unpaid volunteer work with you? complaints and accuses? well done, this clearly supports our motivation to continue working on open source projects. Jun 04 05:30:44 hey zecke Jun 04 05:31:03 koen: hey Jun 04 05:31:48 How to do create a cset wwhen using citool is not possible? Jun 04 05:32:07 bk ci -y"comment" file Jun 04 05:32:10 bk commit Jun 04 05:32:18 bk push Jun 04 05:33:15 bk ci can take multiple files too Jun 04 05:35:21 damn... I messed up Jun 04 05:35:29 sylpheed: some gpgme m4 macros are necessary at configure time. thanks to pb_ for pointing this out. Jun 04 05:35:49 pb_: fixed Jun 04 08:00:23 kaffe: Jun 04 08:00:23 - kaffeh-native: override EXTRA_OECONF instead of appending to it Jun 04 08:00:23 - kaffe.inc: add libffi to DEPENDS Jun 04 08:33:19 * kergoth hugs his new raid controller and 146 gig scsi cheetah Jun 04 08:35:15 kergoth, any closer to figuring out what you want to be when you grow up? Jun 04 08:36:38 not really Jun 04 08:36:53 me neither :) Jun 04 08:37:02 heh :) Jun 04 08:37:23 but at least I can trickle money out the school for doing fun stuff for a while Jun 04 08:37:43 nod Jun 04 08:37:56 * koen suddenly remembers to extend his contract Jun 04 08:38:06 of course if I were to get a real jbo doing what I am doing now, much more $$ Jun 04 08:38:24 Crofton_|laptop: and much more crap from the bossperson Jun 04 08:38:26 and more hassles Jun 04 08:38:29 yep Jun 04 08:39:05 unless i was the boss! Jun 04 08:39:16 in which case I would be my own worst enemy Jun 04 08:41:52 you could give yourself a raise Jun 04 08:41:58 True Jun 04 08:43:23 should I put patches in bugzilla now? I'm going to have some fixes for some stuff in a few days Jun 04 08:44:27 you can always put patches in bugzilla Jun 04 08:44:50 OK, my omni* bb files have some hassles I need to fix Jun 04 08:45:52 morning kergoth Jun 04 09:00:01 hey kergoth Jun 04 09:49:22 do you need a do_stage methd, to put libs etc in STAGINGDIR? Jun 04 09:49:56 yes Jun 04 09:51:10 can anyone tell me why poodles ipkg.conf doesn't have "dests" for /media/card, /media/cf and /home/? Jun 04 09:51:19 is that intentional? Jun 04 09:53:13 probably an oversight Jun 04 09:53:30 though so Jun 04 09:53:37 isn't that /packages ? Jun 04 09:53:40 yep Jun 04 09:56:13 ~beagle Jun 04 09:56:14 methinks beagle is the MasterIA PA-100, an SA1110 based PDA. See http://www.montavista.co.jp/partners/development_tools/master_ia.html, or a dog Jun 04 09:56:36 who is the maintainer for snes9x? is there one? the package maintainer listed is the oe team. the latest snes9x is not rendering anything Jun 04 09:58:59 hmmm dests on tosa still point to /mnt, not /media. /me fixes that, too Jun 04 10:03:12 ~spitz Jun 04 10:03:13 from memory, spitz is the Sharp SL-C3000, or a dog Jun 04 10:03:20 ~akita Jun 04 10:03:22 from memory, akita is the Sharp SL-C1000, or a dog Jun 04 10:03:33 ~coredump Jun 04 10:03:34 extra, extra, read all about it, coredump is the guy who maintains the unofficial OZ3.3.6-pre1 / Opie 1.1.x ROMs at http://www.hentges.net Jun 04 10:03:51 finally, something that's not a dog :) Jun 04 10:03:51 koen: heh 3.3.6 Jun 04 10:04:02 ~koen Jun 04 10:04:03 hmm... koen is our regular image monkey Jun 04 10:04:07 :) Jun 04 10:04:37 <[cc]smart> I would like to make a suggestion to the person giving us the ntp package. You might add these two lines to ntp.conf, to activate local clock as a fallback for system time: Jun 04 10:04:37 <[cc]smart> server 127.127.1.0 Jun 04 10:04:38 <[cc]smart> fudge 127.127.1.0 stratum 14 Jun 04 10:04:53 ibot: no, coredump is the guy who maintains the hentges ROMs @ www.hentges.net Jun 04 10:04:54 CoreDump|home: okay Jun 04 10:05:10 127.127 ? Jun 04 10:05:28 CoreDump|home: I'm fairly sure those aren't ROMs Jun 04 10:05:30 <[cc]smart> it's a symbolic IP address to stick to the same format Jun 04 10:05:49 reenoo_work: not you as well! Jun 04 10:05:52 <[cc]smart> do not interpret it as an IP actually. Jun 04 10:06:01 pb_: :) Jun 04 10:06:02 <[cc]smart> it means: use the local clock. Jun 04 10:06:14 pb_: sorry, couldn't resist ;) Jun 04 10:06:19 EXTRA_IMG_CMD_openzaurus = " -ro" Jun 04 10:06:21 reenoo_work: they can work as true "ROMs" when you boot off of SD or CF :) Jun 04 10:06:22 there Jun 04 10:06:35 ibot: no, coredump is the guy who maintains the hentges distribution @ www.hentges.net Jun 04 10:06:38 okay, CoreDump|home Jun 04 10:06:42 ;) Jun 04 10:08:02 s/ROM/firmware maybe. Should get the point across to newbies Jun 04 10:10:08 kaffe (kaffeh-native): add --without-gmp to EXTRA_OECONF to avoid relying on a native gmp. Jun 04 10:10:40 ROM -> image Jun 04 10:10:53 looks like all Z fstabs got updated to mount a tmpfs at /media/ram. should we allow users to install stuff there? AFAIK only collie has a "real" not-so-volatile ramdisk, right? IMO ipkg dests for /media/ram should be removed on these devices. Jun 04 10:12:35 CoreDump|home: email oe@ Jun 04 10:12:53 hmm will do Jun 04 10:23:17 CoreDump 1.3498 ipkg.conf (tosa): s/mnt/media Jun 04 10:23:35 CoreDump 1.3497 Add propper ipkg destinations for poodle Jun 04 10:43:24 hi Jun 04 10:43:39 hey france Jun 04 10:43:40 does anybody know who is maintaining the old wiki for OE? Jun 04 10:43:45 koen: hi Jun 04 10:43:56 I'm pretty sure that's "no one" Jun 04 10:43:58 it is being filled with spam Jun 04 10:44:21 form links to porn to every prescription drug, I ever heard of. Jun 04 10:47:32 yeah, old wiki is unmaintained, that's part of what makes it an "old wiki" Jun 04 10:47:49 and if it's getting spammed it should probably be kicked into read-only mode or taken offline Jun 04 10:47:52 france: lol Jun 04 10:48:21 is there a nice simple frontend to madplay? Jun 04 10:49:13 keturn: it is usless in readonly mode, as most of the pages have been cleared off by the spammers. Jun 04 10:49:31 keturn: the pages would need to be reverted to a previous version to be useful. Jun 04 10:50:27 keturn: the good pages are now several layers deep, as spammers have squashed the pages of other spammers. Jun 04 10:52:16 hi Jun 04 10:54:37 kaffe (kaffeh-native): add --enable-pure-java-math to really get rid of gmp as a native dependency. Jun 04 10:56:31 whoa... when i boot, the GPE logo is shown, and then half of it gets cut off, the machine dies and the half that remains slowly deteriorates on screen Jun 04 10:56:56 any idea what might be going on? i think something's wrong with the hardware Jun 04 10:57:10 I have followed the OpenEmbedded GettingStarted Wiki page and when I try to bitbake something I get a patch error: 6 chunks of 7 failed at package gnu-config-native-0.1cvs20050331-r3 Jun 04 10:58:01 i want to start develop something for gpe but i'm stuck at the very first step... Jun 04 10:59:05 DarkMan|: looks like a too old CVSDATE Jun 04 10:59:16 ok, i'll check that Jun 04 10:59:16 thx Jun 04 10:59:39 hrm Jun 04 10:59:51 CVSDATE_gnu-config-native = "20050602" should do Jun 04 11:00:17 just don't set a global cvsdate Jun 04 11:02:34 has anyone seen the symptoms i have described? Jun 04 11:02:39 CVSDATE is set at familiar-0.8.2.conf to 20050331 Jun 04 11:02:47 so I don't touch that, right? Jun 04 11:03:08 in your conf/local.conf you just change the date for one package Jun 04 11:03:13 ok Jun 04 11:03:14 zecke: no Jun 04 11:03:22 koen: what no? Jun 04 11:03:24 don't try to build old distros with a new repo Jun 04 11:03:29 i have a feeling something's wrong with the flash... Jun 04 11:03:40 miraage: topic, 3rd item Jun 04 11:03:41 koen: that is a different issue Jun 04 11:03:57 zecke: that's the real problem, gnuconfig is only a symptom Jun 04 11:03:58 DarkMan|: either check out openembedded with the familiar-0.8.2 tag Jun 04 11:04:12 DarkMan|: or use familiar-0.9 Jun 04 11:04:24 0.9.0 even :) Jun 04 11:04:31 ok Jun 04 11:04:43 the packages will be library compatible to familiar-0.8.2? Jun 04 11:05:05 depends on the library Jun 04 11:05:10 but generally, no Jun 04 11:05:41 i have too little flashs space in my 3630 to install duplicate libraries... Jun 04 11:06:32 so I need to build a full familiar-0.9 to have all packages linked with the same libraries? Jun 04 11:07:10 that or check out an OE tree with the familiar-0.8.2 tag Jun 04 11:07:29 ok thanks Jun 04 11:14:45 chanserv: fine, be that way. Jun 04 11:16:28 zecke: you said something about the ;md5= last night, did you say it was possible, or that you implemented it? Jun 04 11:20:47 * france is away: Away Jun 04 11:24:16 koen: possible Jun 04 11:24:33 koen: if you start adding one ;md5= Jun 04 11:24:38 koen: I'll immediately implement it Jun 04 11:25:08 * zecke wants a package of Vla from Albertijn (is there a 'h' missing?)! Jun 04 11:25:21 yes, and a 'e' :) Jun 04 11:25:33 Albert Heijn Jun 04 11:25:38 lol Jun 04 11:25:39 My ex-employer :) Jun 04 11:25:53 * zecke imagines that Jun 04 11:26:02 Koen working at Alber Heijn in venlo ;) Jun 04 11:26:13 heh Jun 04 11:27:26 I see a problem Jun 04 11:27:28 .incs Jun 04 11:27:41 koen: right Jun 04 11:27:51 koen: new key? Jun 04 11:28:20 MD5_SUM_foobar = "abc" Jun 04 11:28:38 hwo does that handle multiple sources? Jun 04 11:29:05 koen: using overides Jun 04 11:29:14 MD5_SUM_foobar.tar.bz2 = "abc" Jun 04 11:29:21 (not sure the dots are allowed) Jun 04 11:29:31 MD5_SUM_gruft.tar = "bce" Jun 04 11:30:06 can we enumerate SRC_URI 'members'? Jun 04 11:31:11 koen|tv: no not too good Jun 04 11:31:21 koen|tv: fetch.py does it though Jun 04 11:31:53 we could only do the first item Jun 04 11:31:59 MD5="33425234" Jun 04 11:36:28 koen|tv: also with inc one could do SRC_URI = "http://foo.foo.foo;md5sum=${SUM_1}" Jun 04 11:45:35 zecke: yes, but if you abstract the SRC_URI that falls apart Jun 04 11:46:28 koen|tv: in the file your including the 'base' you'Ve to specifiy SUM_1 Jun 04 11:46:39 not too nice though Jun 04 11:52:27 me again... I checkout a tree for familiar-0.8.2 and now the patching works, but it tries to decompress quilt-native-0.39 without downloading, and it stops. The strange thing is that before shows: NOTE: package quilt-native-0.39-r0: task do_fetch: completed Jun 04 11:52:51 but the tar.gz isn't in my sources dir Jun 04 11:53:28 DarkMan|: hmm Jun 04 11:53:49 DarkMan|: wipe your tmp directory Jun 04 11:56:47 DarkMan|: maybe removing tmp/stamps/quilt* will help Jun 04 11:57:38 removed full tmp and trying again... Jun 04 12:04:16 samba distributes source tarballs by bittorrent now. could add a fetch method for that. as they point out, it'd be a good idea to check the signature, though. Jun 04 12:04:55 I get the same error removing tmp directory Jun 04 12:05:20 where stores bitbake the information for downloaded sources? Jun 04 12:07:27 DarkMan|: in the DL_DIR Jun 04 12:08:16 not the tarballs, i mean, if saves somewhere what packages has downloaded before Jun 04 12:11:49 DarkMan|: which version of bitbake do you've Jun 04 12:11:55 DarkMan|: how does your BBPATH look? Jun 04 12:16:31 bitbake version... the one retrieved with "svn co svn://svn.berlios.de/bitbake/trunk/bitbake" not installed but added to path, and BBPATH="/home/darkman/ipaq/familiar/build:/home/darkman/ipaq/familiar/openembedded" Jun 04 12:26:11 starting setup all over... Jun 04 12:26:48 DarkMan|: bitbake -v Jun 04 12:27:04 verbose? Jun 04 12:27:22 or you mean --version Jun 04 12:28:14 DarkMan| version :} Jun 04 12:28:40 BitBake Build Tool Core version 1.3.0, bitbake version 1.3.1 Jun 04 12:30:10 hmm looks good as well Jun 04 12:30:12 zecke: so ;md5= for single.bbs and ;md5=${SUM1} + SUM1 = "" for includers + includees Jun 04 12:30:48 koen|tv: why the ${S}+S1= "" Jun 04 12:31:08 the first goes in the inc, the second in the .bb Jun 04 12:31:36 alternatively we could use overrides Jun 04 12:31:43 MD5_SUM_filename = "SUM" Jun 04 12:32:31 that would solve both situations Jun 04 12:32:49 koen|tv: I don't know how filename would like ;) Jun 04 12:33:20 MD5_SUM_qt-embedded-${PV}-free.tar.gz = "1f7ad30113afc500cab7f5b2f4dec0d7" Jun 04 12:34:03 koen|tv: I don't know fi we can have '.' there but probably we can Jun 04 12:34:38 OT: www.rockpalast.de/livestream Jun 04 12:35:17 yeah, I think "." is allowed in variable names. Jun 04 12:35:55 would anyone veto a scp fetcher? Jun 04 12:36:42 no, I think an scp fetcher would be a fine idea Jun 04 12:39:03 no way here... it downloads thcrut-1.2.5.tar.gz but not finds quilt_savannah.nongnu.org_VER_0_39_.tar.gz Jun 04 12:41:47 DarkMan|: ok someone on the ML has the same problem Jun 04 12:50:34 and has found any solution? Jun 04 12:51:57 can I see that message? the maillist archive is from oct-2004 Jun 04 13:02:43 koen|tv: let us finalize on MD5_SUM Jun 04 13:18:53 good plan Jun 04 13:33:11 hi Jun 04 13:34:06 that md5 will look strange... first SRC_URI with list and then lines with MD5 for each entry? Jun 04 13:34:08 hey hrw Jun 04 13:34:34 hrw: proposals are welcome Jun 04 13:34:55 the stupid thing is you need one MD5SUM for each downloaded file Jun 04 13:35:45 SRC_URI="http://example.net/file.tar.bz2;md5=4273482742947239472394" Jun 04 13:36:23 that was the plan, but that doesn't owrk with abstractes SRC_URI's Jun 04 13:36:34 (.inc, bbclasses) Jun 04 13:37:51 so maybe if there is no "md5" part in SRC_URI then use MD5SUM_file option? Jun 04 13:38:09 that's what we've thought of so far Jun 04 13:38:59 ok Jun 04 13:44:02 DarkMan|: could check gmane.org Jun 04 13:45:13 hrm. I defined do_srec. It gets included in in the 'run.image_cmd.nnn' scripts. I said 'addtask srec after do_rootfs'. but I don't think anything is invoking do_srec. Jun 04 13:47:25 keturn: if you define do_srec you'll also want 'addtask do_srec ...' Jun 04 13:50:08 that'd make some kind of sense, but everything I see in classes/ is defined with "do_" and the do_ is never included in the first arg to addtask Jun 04 13:50:12 * keturn tries it, though. Jun 04 13:51:06 heh, right. sorry Jun 04 13:51:22 ~lart inconsistencies Jun 04 13:51:22 * ibot strangles inconsistencies with a 9-pole serial cable Jun 04 13:52:04 look into packages/kdepimpi/kdepimpi-base.inc - I have do_unstage() there Jun 04 13:55:09 Is there some command to reverse a patch file? Jun 04 13:55:20 Or some way to tell oe to applyu it reversed? Jun 04 13:55:36 there's a reverse flag to 'patch', which might be -R Jun 04 13:56:26 That's not quite what I need... Jun 04 13:57:00 kind of SRC_URI_append="file://my.patch;patch=1;reverse=1"? Jun 04 13:57:37 hrw: Yes, is that valid syntax? Jun 04 13:57:41 RP: no Jun 04 13:58:01 RP: but can be easily added I think.. we already have "pnum" Jun 04 13:58:40 I'll feed them in by hand for now. It would be a nice feature to add though Jun 04 13:59:08 I'm amused Jun 04 13:59:15 Oh bitbake uses so much memory Jun 04 13:59:27 Oh the new bitbake does not work with stuff prior to cset... Jun 04 13:59:43 zecke: you're easily amused. Jun 04 13:59:51 I guess flexibility is good and all, but, um, I'd hate to have patches like that in my codebase. Some patches get applied one way and others in reverse? Sux. Jun 04 14:00:24 just in terms of keeping things readable and reducing surprises. Jun 04 14:00:44 yeah, it does seem a bit bogus to have reversed patches in the works. Jun 04 14:01:10 it doesn't seem that unreasonable to expect people to turn their patches the right way round before checking them in. Jun 04 14:01:13 pb_: now that I care less ;) Jun 04 14:01:18 pb_: I can be easily amused Jun 04 14:01:32 heh Jun 04 14:01:50 I'm just trying to test something and way to revert some changes in a patch. Its not going into oe... Jun 04 14:02:09 I just can't find an easy way to do this. quilt has no option to import a diff reversed :-/ Jun 04 14:02:24 RP: luckily Jun 04 14:02:59 zecke: not really. How do I apply this reversed diff against my code base in any sane way... Jun 04 14:03:32 RP: hmm do not apply the patch in the first place Jun 04 14:03:37 I'll just have to use patch which leaves me no way to generate a sane diff later... Jun 04 14:03:41 RP: use overrides to apply Jun 04 14:04:16 hrw: can I talk you into looking at HTML generation in your spare time? Jun 04 14:04:24 Its only a partial reversal I'm trying... Jun 04 14:04:59 load the patch in emacs and hit M-R a lot, which will swap all the +s and -s around Jun 04 14:05:24 That's not the point :) Jun 04 14:06:24 zecke: ok Jun 04 14:06:30 RP: I don't really understand what you're saying. What's the big deal about creating a patch that backs out the changes you are trying to revert? Jun 04 14:06:54 hrw: I hope to finish bitdoc tonight (generating documentation out of documentation.conf) Jun 04 14:07:39 zecke: with wikistyle autohighlighting? Jun 04 14:08:07 pb_: I just can't find an easy approach to make this particular opertation Jun 04 14:09:15 zecke: I still haven't found an easy way for packages.openembedded.org Jun 04 14:09:30 koen|tv: with crappy non conformant, vendor locked in, HTML Jun 04 14:09:40 koen|tv: use bitbake to get the information ;) Jun 04 14:10:35 p.oe.o should have unmodified info + parsed info Jun 04 14:11:12 zecke: so you want it to create static pages for all recipes? Jun 04 14:11:15 koen|tv: tinderbox data to see if the package built for the architecture Jun 04 14:11:32 hrw: no describing the bbclasses (groups) and the keys Jun 04 14:11:55 hrw: just some easy HTML generation, I just want you to take a look when I've checked it in Jun 04 14:13:13 zecke: ok - can look into Jun 04 14:14:10 OT - it looks like in some time I will have to learn lot about iptables or will switch all services to non-standard ports.. Jun 04 14:14:11 happydoc worked well for eariler versions .. Jun 04 14:18:32 * france is back (gone 02:57:44) Jun 04 14:22:30 hrw: why is that? Jun 04 14:24:43 keturn: my sshd allow to connect for one user only and is flooded by dictionary 'attacks' (which is reported by logcheck) Jun 04 14:25:53 does the one user always come from the same IP? Jun 04 14:29:12 keturn: it is one ip per attack and use usernames from dictionary. Jun 04 14:30:34 I mean the valid user. If so it's pretty trivial to make a whitelist with iptables. but if you want to log in from whatever random cafe you happen to be at, iptables is less helpful. Jun 04 14:34:05 I found whitelisting for that kind of 'attack' - will look does it works Jun 04 14:35:54 ok, my problem with "addtask srec after do_rootfs" seemed to go away when I changed it to "addtask srec after do_rootfs before do_build" Jun 04 14:47:19 What's the recommended thing for a task to do if some necessary variable isn't set? Fail silently? Fail with a warning? Throw an exception? Execute commands anyway and have them break horribly? Jun 04 14:47:40 throw an Exception Jun 04 14:49:17 how does that work in not-python? Jun 04 14:51:19 I could wrap the not-python function in a python function just for the purpose of doing that, I suppose, but that hurts. Jun 04 14:55:21 cu all Jun 04 14:55:47 'night hrw|gone Jun 04 14:59:46 * keturn finds the oewarn, oefatal functions defined in base.bbclass. Jun 04 15:37:13 oops. Something I did is making do_rootfs get called twice. Jun 04 15:40:03 keturn: hehe, when I wrote my test.bbclass I had the same problems Jun 04 15:40:22 keturn: you miss some src_ in front of one of your tasks Jun 04 15:41:15 huh? Jun 04 15:43:05 srec_ Jun 04 15:43:21 you've stuff like addtask foo after fetch Jun 04 15:43:30 srec_do_foo() { Jun 04 15:43:30 } Jun 04 15:43:44 I had some weird issues when forgetting the srec_ prefix Jun 04 15:46:27 blurrrr... Jun 04 15:46:55 keturn: yes the bbclasses are pure vodoo Jun 04 15:48:31 omg my eyes Jun 04 15:48:37 * keturn opened BBHandler.py Jun 04 15:48:45 hehe Jun 04 15:48:51 wasn't me Jun 04 15:52:35 some native english speaker around? Jun 04 15:54:08 err, not native, but close enough ;) Jun 04 15:55:24 Twiun: could you review some abstract? Jun 04 15:55:25 zecke: fire away Jun 04 15:55:30 sure Jun 04 15:55:31 just ~250 words Jun 04 15:57:33 I've mailed drw for now Jun 04 15:58:23 'k Jun 04 15:58:43 Twiun: what is your email adress? Jun 04 15:58:52 zecke: steph AT tangency.co.uk Jun 04 16:04:12 you can send to /me @ /me . net too. it'll give me something to read while I wait for the bb cache to reparse. Jun 04 16:05:09 for that it is too bad Jun 04 16:08:32 err, can OOo open kword files? Jun 04 16:08:36 * Twiun is stuck on win32 Jun 04 16:08:58 hehe no it can not :} Jun 04 16:09:10 damn. Jun 04 16:09:30 that's stupid. they should be able to open each others files ;) Jun 04 16:09:44 esp as they're both open source Jun 04 16:10:12 Twiun: with OASIS this is the case Jun 04 16:32:15 nope, naming it srec_do_srec still has the double do_rootfs Jun 04 16:36:48 can I get at the data dict from bbshell? Jun 04 16:43:19 I've killed my collie Jun 04 16:43:26 keturn: probably with expert mode Jun 04 16:52:58 *puh* it boots again Jun 04 16:59:03 ~lart bad assumptions Jun 04 16:59:03 * ibot flings poo at bad assumptions Jun 04 17:00:33 it kind of sucks when you find out that you have to rewrite half of the codebase of a project because someone decided it would be a good idea to give a meaning to the order of (key, value) pairs :/ Jun 04 17:01:54 what's the context? Jun 04 17:04:01 as part of a university project we use some sort of product database. products have a set of attributes (att_name, att_value). Jun 04 17:04:54 the attributes were implemented as a Vector (wasn't me) Jun 04 17:05:01 and the order matters Jun 04 17:05:27 urgh. Jun 04 17:05:33 * reenoo_ nods Jun 04 17:05:56 why didn't they use hashtables??? oh well Jun 04 17:06:07 you could write a Vector wrapper that swaps them ;) Jun 04 17:06:25 I told them to use Maps. but they didn't. Jun 04 17:06:44 some people find the concept of hashes difficult to grasp :( Jun 04 17:07:57 well, with the consequence that now performance sucks for obvious reasons :/ Jun 04 17:08:24 * Twiun nods Jun 04 17:58:30 * reenoo_ heads off to bed for now Jun 04 17:58:33 'night all Jun 04 18:28:45 okay kids, it took me bloody long enough to see it, but task_cache is br0ken Jun 04 19:08:59 argh. my explanation for why it is br0ken does not quite hold water. Jun 04 19:18:14 ah. new explanation. Jun 04 19:42:16 bug #60. one-line patch. Jun 04 19:49:13 can i get the patches that oe uses without having to install oe somehoe? Jun 04 21:02:42 keturn: ? Jun 04 21:44:11 http://www.wkbt.com/Global/story.asp?S=3416199 Jun 04 21:44:33 "For those projects that will not avail themselves of BitKeeper's commercial program there is an export utility that will convert a BitKeeper repository into a CVS repository." Jun 04 22:16:28 we know about it Jun 04 22:16:37 there are no plans to switch to cvs. Jun 04 22:43:01 ggilbert__: It may be the only exit from BK-format... convert to CVS, then to something else Jun 04 22:48:11 Luke-Jr, think about taht for a second ... Jun 04 22:48:36 there is nothing to stop fromconverting a local tree into a new repository type Jun 04 22:49:27 all the bk history is lost, but then again history with CVS is almost unmaintainible Jun 04 22:51:25 does that mean that the free bk-client will also stop working? Jun 04 22:52:47 it never worked in the first place and is actually one of the major causes for BK closing down openprojects Jun 04 22:53:11 I just used it tonight... Jun 04 22:53:29 to check OE metadata Jun 04 22:53:55 update bk://openembedded.bkbits.net/openembedded ~/bb/openembedded Jun 04 22:54:23 emte: I'm not talking about sourcepuller Jun 04 22:54:33 I'm talking about the one bitmover released Jun 04 22:54:57 @ http://www.bitmover.com/bk-client.shar Jun 04 22:55:19 jumpkick: does it matter if noone can commit? Jun 04 22:55:48 Luke-Jr: it's both ways... Jun 04 22:55:56 if the devs get keys to commit Jun 04 22:56:36 but no non-devs can checkout without a key... then it's another barrier to growing the dev pull for OE Jun 04 22:56:44 dev pool Jun 04 22:57:06 jumpkick: using BK alone is a barrier Jun 04 22:57:27 Luke-Jr: you'll get no disagreement from me on that Jun 04 22:58:10 Luke-Jr, you have been told repeatedly why we were using bk Jun 04 22:58:21 there is no other usable tool Jun 04 22:58:46 indulge me... what's wrong with svn or monotone or arch? Jun 04 22:58:48 emte: It doesn't matter, it's still a barrier Jun 04 22:59:05 monoton and arch werent usable Jun 04 22:59:06 emte: you *could* forego the few technical benefits of BK, no? Jun 04 22:59:15 monotone is actually a candidate IF it worked properly Jun 04 22:59:24 svn was the wrong architechture for how people worked Jun 04 22:59:35 I actually haven't seen anything BK can do that Svn cannot Jun 04 22:59:44 or at least, that OE has used Jun 04 22:59:44 then thats your problem. Jun 04 22:59:52 Luke-Jr: work offline Jun 04 22:59:57 for an extended period of time Jun 04 22:59:57 svn cant handle multiple clients at the same time Jun 04 23:00:03 ggilbert__: Svn has extentions for that, I think Jun 04 23:00:21 yes, its a hack on top of a system thats architected in a way that doesnt fit our needs. Jun 04 23:00:26 heya kergoth Jun 04 23:00:29 hey Jun 04 23:00:38 you finally moved in yet? Jun 04 23:00:42 yeah Jun 04 23:00:48 all settled Jun 04 23:00:48 good good Jun 04 23:01:08 now we just need to find ya a pretty redneck gf :P Jun 04 23:01:28 kergoth: and those 'needs' are more important than the benefits of using another tool? Jun 04 23:01:52 Luke-Jr, when you have a few hundred users, yes Jun 04 23:02:30 Luke-Jr: "benefits"? i dont call using hte wrong tool for the job a benefit. Jun 04 23:03:08 anyway, i'm tired of wasting time explaining shit thats been explained on countless occasions. we use bk because its the right tool for the job, and we'll switch to the best available alternative when necessary Jun 04 23:03:14 a crescent wrench works as a hammer but not very long Jun 04 23:04:02 i knew someone that used duct tape and glue to fix leaky fuel injectors Jun 04 23:04:13 lol Jun 04 23:04:32 at least there wasnt fuel pooling up on the engine anymore Jun 04 23:04:39 :) Jun 04 23:05:36 emte: there are a few hundred users? wow.... Jun 04 23:06:09 nevermind... Jun 04 23:06:13 I'm too tired Jun 04 23:06:30 yes Jun 04 23:06:51 the people in this channel as a shadow of the user pool Jun 04 23:06:57 are Jun 04 23:06:59 * Jun 04 23:07:27 is patcher built by bitbake? Jun 04 23:08:02 not really Jun 04 23:08:05 patcher isnt built Jun 04 23:08:07 I had to put in INHERIT += patcher, cause quilt doesn't like me tonight Jun 04 23:08:19 it's a perl script Jun 04 23:08:25 oh... okay Jun 04 23:08:42 I was worried for a sec, cause I don't have it in my path **** ENDING LOGGING AT Sat Jun 04 23:59:56 2005