**** BEGIN LOGGING AT Sun Dec 30 02:59:59 2012 Dec 30 03:12:52 build #136 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/136 Dec 30 05:29:20 build #102 of iop32x is complete: Failure [failed shell_11] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/102 Dec 30 08:24:45 juhosg r34934 trunk/target/linux/ramips/files/arch/mips/ralink/rt305x/mach-mzk-w300nh2.c * ramips: rt305x: fix garbled name in mach-mzk-w300nh2.c Dec 30 08:35:33 build #99 of gemini is complete: Failure [failed shell_11] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/99 Dec 30 12:38:28 build #94 of octeon is complete: Failure [failed shell_11] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/94 Dec 30 12:39:37 blogic r34935 trunk/target/linux/lantiq/image/ARV4518PW.dts * [lantiq] fixes usb port on arv 4518 Dec 30 13:21:07 I'm getting some queries over on #openwrt as to the reasoning behind my ddns-scripts wget patch. Dec 30 13:21:48 Try this command to see: /bin/busybox wget http://an.example.org:password@dyn.dns.he.net/nic/update?hostname=an.example.org&myip=10.1.1.1 Dec 30 13:22:15 Now strip /bin/busybox and use GNU Wget and watch it fail. Dec 30 13:31:29 try only wget 'http://an.example.org:password@dyn.dns.he.net/nic/update?hostname=an.example.org&myip=10.1.1.1' Dec 30 13:40:35 luka: busybox wget gets a "badauth" response, gnu wget only a 401 and retries again and again. interestingly, if you specify the credentials with --user/--password; gnu wget also manages to retrieve the "badauth" Dec 30 13:41:01 so it seems there is a difference in behaviour between --user/--password and userpart in url Dec 30 13:52:06 jmccrohan: adding "--no-http-keep-alive" for gnu wget helps, too Dec 30 14:04:49 jmccrohan: you should be able to detect whether it's gnu or busybox wget by testing if /usr/bin/wget is a symlink (-L), then add "--no-http-alive" if it isn't to the wget call - should make it compatible with both :) Dec 30 14:59:33 nbd r34936 trunk/package/uci/Makefile * uci: disable parallel builds again, the lua<->libuci dependency is not safe Dec 30 15:29:24 hauke r34937 trunk/package/acx-mac80211/Makefile * acx-mac80211: fix compilation Dec 30 15:33:49 hauke r34938 trunk/package/system/rtc-rv5c386a/src/rtc.c * rtc-rv5c386a: fix compilation for kernel >= 3.4 Dec 30 16:13:55 KanjiMonster: Thanks for that. Digging into GNU Wget source to see what the logic is there before doing anything here. :) Dec 30 16:15:20 jmccrohan: the problem seems to be that the he.net server does not handle a second request with keep alive properly; the busybox wget probably does not use keep alive at all Dec 30 16:25:41 KanjiMonster: Indeed. I thought the server was running standard apache, but that it only the root domain. Dec 30 16:25:58 The cgi script returns the following Server: dns.he.net v0.0.1 Dec 30 16:30:06 jmccrohan: i think the best course of action is just to have the script specify "busybox wget" Dec 30 16:31:27 KanjiMonster: Apply as it then? It shouldn't add any bloat as busybox wget and the required options are default=y anyway. Dec 30 16:32:13 we've had the same problem appear in the boot sequence when the non-busybox "mount" is present Dec 30 16:32:49 jmccrohan: I'd rather make it work with gnu wget, too Dec 30 16:33:14 KanjiMonster: unnecessary bloat Dec 30 16:33:18 else you can't make it work with ssl urls by installing wget-openssl Dec 30 16:34:03 over-engineering small systems is bad, yo Dec 30 16:34:15 DonkeyHotei: it's like two lines Dec 30 16:35:07 it's not needed Dec 30 16:35:29 DonkeyHotei: it's more bloat enforcing busybox wget to use it Dec 30 16:35:48 two lines here, two lines there, everywhere a two lines -- this is what moore's law is all about Dec 30 16:36:05 how is busybox more bloat? Dec 30 16:37:00 KanjiMonster: TBF, busybox wget is default, and if they have enough space for wget-ssl, then I don't think space is a concern. Dec 30 16:38:14 jmccrohan: my issue is that your patch is hardcoding busybox wget; and to use the ssl wget you will need to modify the dyndns scripts Dec 30 16:38:15 the issue comes up when the space for wget-ssl is not there Dec 30 16:38:47 why would you EVER need to use the ssl wget for ddns? Dec 30 16:39:19 DonkeyHotei: because you might not want to transfer your ddns passwords (even hashed) unencrypted? Dec 30 16:39:51 tell me what ddns service supports ssl. afaik, there are none Dec 30 16:40:10 he.dns does, but uses a self-signed cert Dec 30 16:40:27 DonkeyHotei: dyn.com does Dec 30 16:40:49 ddns passwords should be throwaways Dec 30 16:41:09 KanjiMonster: Doesn't ddns-scripts use CURL for HTTPS? http://git.dereenigne.org/openwrt-packages.git/blob/refs/heads/trunk:/net/ddns-scripts/files/usr/lib/ddns/dynamic_dns_updater.sh#l101 Dec 30 16:41:49 jmccrohan: indeed it does. Dec 30 16:43:07 jmccrohan: anyway, [ -L "/usr/bin/wget" ] || retrieve_prog="/usr/bin/wget --no-http-keep-alive -O - "; should be enough to keep it compatible to the full wget Dec 30 16:46:51 I guess. I want full Wget installed, but I'd rather the busybox wget called by ddns-scripts. Maybe a uci config option if you want full wget might be the best approach? Dec 30 16:48:24 if ssl uses curl, what is the use case for using gnu wget in the script at all? Dec 30 16:48:35 KanjiMonster: ^ Dec 30 16:51:15 jmccrohan: failing an answer to that, i'm back to agreeing with your initial solution Dec 30 16:51:23 jmccrohan: hm, I think that would be too much. Dec 30 16:51:48 right Dec 30 16:52:26 DonkeyHotei: asked the other way round, what's the use case of using the busybox wget if a full blown wget is installed? Dec 30 16:52:56 Speed. Dec 30 16:52:59 because it's known to work reliably with no fuss Dec 30 17:00:11 Delboy: are you around by any chance? Dec 30 17:38:49 flower power enterprises mit gammligem alten openwrt .. :D Dec 30 17:38:50 rofl Dec 30 18:36:46 jmccrohan: so cnt you just pass the --no-check-ssl Dec 30 18:38:18 or --no-check-certificate Dec 30 18:56:45 phaidros: where ' Dec 30 19:06:34 blogic: beim security nightmares talk .. war funny :) Dec 30 20:47:51 build #130 of lantiq is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq/builds/130 Dec 30 21:26:24 phaidros: hmmm its not on youtube yet Dec 30 21:29:43 blogic: btw, -^ lantiq doesn't build because of missing kernel symbols Dec 30 21:30:18 interesting Dec 30 21:32:07 blogic r34939 trunk/target/linux/lantiq/config-3.7 * [lantiq] adds missing symbol Dec 30 22:30:11 blogic: https://www.youtube.com/watch?v=QRLeHPRnu3Q Dec 30 22:32:59 ab 1:17:00 lifecycle management Dec 31 00:14:55 build #125 of rb532 is complete: Failure [failed shell_11] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/125 Dec 31 00:18:20 build #125 of ppc44x is complete: Failure [failed shell_11] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/125 Dec 31 01:16:40 build #137 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/137 Dec 31 02:52:06 build #115 of sibyte is complete: Failure [failed shell_11] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/115 **** ENDING LOGGING AT Mon Dec 31 02:59:59 2012