**** BEGIN LOGGING AT Wed Mar 07 02:59:57 2007 Mar 07 05:59:07 hey andersee Mar 07 06:01:15 nbd: howdy Mar 07 06:01:25 * andersee is mucking about with his network Mar 07 06:01:43 just wanted to let you know that i'm doing something really crazy - building an embedded web framework out of busybox' awk interpreter Mar 07 06:01:52 nbd: heheheh Mar 07 06:02:07 i added a few hooks to it so that i can register c functions Mar 07 06:02:29 and wrote a new source file that will handle cgi stuff and wrap around awk in a mostly nonintrusive way Mar 07 06:02:51 nbd: might as well add a full implementation of CORBA while you are at it. Mar 07 06:02:53 ;-) Mar 07 06:02:55 it'll do some stripped down mvc style Mar 07 06:02:58 i assume you made another multicall target? Mar 07 06:02:59 and it'll be really tiny Mar 07 06:03:04 lst: yes Mar 07 06:03:23 sounds pretty cool Mar 07 06:03:36 i wrote some notes on it: http://nbd.name/webif_tmp.txt Mar 07 06:03:38 nbd: webif_tmp might be of interst Mar 07 06:03:46 damn, pasted while i type. Mar 07 06:03:46 and a large part of what is described there is actually implemented right now Mar 07 06:03:48 :) Mar 07 06:03:56 I was very impressed when I saw the awk based openwrt web management interface Mar 07 06:04:36 the main problems are that it's still too slow and uses too much memory, and i want to make it a lot easier to write web interface pages Mar 07 06:04:47 andersee: felix likes abusing things to do what they werent meant to do.. :) Mar 07 06:04:50 all the forking in the shell is adding a lot of overhead Mar 07 06:04:56 sure -- all that exec'ing and such has a lot of overhead Mar 07 06:05:10 so i'm doing pure awk stuff with a minimal ssi parser Mar 07 06:05:25 and i want the bulk of the awk code to be preloaded with the http server :) Mar 07 06:05:28 that's the next step Mar 07 06:07:04 nbd: looks pretty slick Mar 07 06:07:27 maybe i can submit some of it back to busybox when it's done Mar 07 06:07:38 it could be used for a lot of interesting stuff Mar 07 07:32:01 anybody here running Ubuntu Edgy or newer? (where /bin/sh defaults to dash) Mar 07 07:33:22 yes Mar 07 07:33:24 or Debian for that matter. Mar 07 07:33:31 dpkg-reconfigure -plow dash Mar 07 07:34:15 it is good for catching bugs. Mar 07 07:36:03 yes it is, but I've noticed that some packages were written to expect the behavior of bash and/or the coreutils echo Mar 07 07:36:22 murb: have you tried compiling amwall? Mar 07 07:37:00 I'm building all the packages to try and work out any kinks before I turn this box into my default build box. Mar 07 07:47:37 hmm the chroot i use for building openwrt is actually has /bin/sh as bash. Mar 07 07:47:41 i can change that though. Mar 07 07:48:19 that shouldn't matter, I have set my env to use bash also... but it still doesn't appear to work Mar 07 07:49:32 oh now this happens: Please do not compile as root. Mar 07 07:49:39 when i use fakeroot make Mar 07 07:49:58 murb: yeah, nbd added that check quite a while ago. Mar 07 07:49:59 (which i only use because things used to break horribly like wrong permissions) Mar 07 07:50:17 s/permissions/ownership/ Mar 07 07:50:26 i'll see if that still happens and open a bug if. Mar 07 07:50:52 oh, I see what I issue is... Mar 07 07:51:42 I merely symlinked ~/bin/sh to /bin/bash thinking that would fix it, but it is hardcoded to /bin/sh Mar 07 07:52:19 but still, quite annoying that amwall's Makefile expects bash's echo behavior Mar 07 07:52:41 dash's internal echo doesn't recognize the -e arg Mar 07 07:53:41 it always handles escaped chars, where-as the default in coreutils' echo and bash's echo is to default to -E Mar 07 08:02:25 <[mbm]> dash? ugh Mar 07 08:04:06 [mbm]: yeah. Mar 07 08:04:53 <[mbm]> ubuntu? Mar 07 08:04:59 [mbm]: yes. Mar 07 08:05:04 <[mbm]> *sigh* Mar 07 08:05:15 Started with Edgy /bin/sh -> /bin/dash Mar 07 08:05:23 <[mbm]> yeah I know the pain Mar 07 08:05:28 this is true of etch if you have dash installed. Mar 07 08:05:31 <[mbm]> I always swapped it back Mar 07 08:05:52 [mbm]: assuming /bin/sh is bash is a bug. Mar 07 08:06:07 [mbm]: I swapped it back too, but I wanted to confirm the broken-ness Mar 07 08:06:12 <[mbm]> murb: yeah I know Mar 07 08:06:33 murb: if I didn't know better I would swear you took that from the Ubuntu "bug" on the issue Mar 07 08:06:46 Bartman007: i'm running etch Mar 07 08:06:54 murb: dash is a bug itself Mar 07 08:07:03 yeah, but those are the exact words from the bug Mar 07 08:07:05 * [mbm] runs fiesty fucker Mar 07 08:07:08 Kaloz: thank you! Mar 07 08:07:34 [mbm]: I run that on most, but this is a friend's machine, so I'm not going to upgrade without him knowing Mar 07 08:07:35 murb: if they needed something faster then bash, they could have go with busybox' ash Mar 07 08:07:51 Kaloz: yeah, cool complete userland in busybox :-) Mar 07 08:08:02 ooh, that reminds me... Mar 07 08:08:03 but it doesn't have an X server yet! Mar 07 08:08:13 murb: that's my plan, as i'm sick of the bloat and crap Mar 07 08:08:15 <[mbm]> Kaloz: but ash doesn't SOUND as fast as dash! Mar 07 08:08:28 murb: and yeah, [mbm] doesn't like my openwrt "desktop" idea :P Mar 07 08:08:32 [mbm]: very good point! Mar 07 08:08:50 Kaloz: you'll have to aquire one of those usb framebuffer cards. Mar 07 08:08:55 hahaha, dash is broken, even busybox ash handles echo -e properly Mar 07 08:09:09 murb: nah, i mena as real desktop alternative, for exmaple my notebook Mar 07 08:09:28 <[mbm]> Kaloz: olive? Mar 07 08:09:30 Bartman007: it has "d" as debian, it can't be wrong Mar 07 08:09:38 Bartman007: maybe it's a random feature Mar 07 08:09:50 Kaloz: yeah, I'm afraid to file a bug, because I dont' want to be slaughtered by devs Mar 07 08:09:53 [mbm]: not really my tasta, and still not that small and fast as i want Mar 07 08:10:07 taste* Mar 07 08:10:19 Kaloz: btw, I was going to try porting Xserver soon Mar 07 08:10:24 [mbm]: but right, isteve has some good ideas Mar 07 08:10:34 (the evolution of kdrive) Mar 07 08:10:34 [mbm]: i know that guy for a few years over ircnet Mar 07 11:33:42 heh, Debian insists that dash operates properly and coreutils echo is broken... http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=401481 Mar 07 11:35:56 <[mbm]> there's two ways to deal with standards Mar 07 11:36:07 <[mbm]> 1> that isn't in the standards; I refuse to support it Mar 07 11:36:41 <[mbm]> 2> I have a cool new feature and I'm going to push it as a standard until it actually is Mar 07 11:37:31 <[mbm]> unix tends towards #1 and microsoft stuff tends to be #2 Mar 07 11:38:47 hcg * r6534 /trunk/target/linux/at91-2.6/ (config/default patches/003-gpio-driver.patch): Added GPIO driver Mar 07 11:38:50 Bartman007: They're not claiming that coreutils echo is broken, merely that it's above and beyond the standard, and you need to ensure you're in a non-standard environment before you use non-standard extensions. Mar 07 11:39:03 whether Debian is in the wrong or not I thing they should add the one/two lines of code needed to ignore an initial non-escaped -e Mar 07 11:39:40 <[mbm]> that would be admitting the standard sucks Mar 07 11:40:39 here's how I see it. The project is Debian GNU/Linux; Linux kernel, GNU userland. Coreutils is part of the GNU userland; therefore Debian should stfu and conform to their name. Mar 07 11:40:56 but that makes too much since so forget it Mar 07 11:41:07 <[mbm]> wouldn't that be linux/gnu? Mar 07 11:41:24 * [mbm] wonders how the gnu part got first Mar 07 11:41:25 [mbm]: order isn't important... Mar 07 11:42:00 [mbm]: RMS took a page from Balmer's book and threw a chair. Mar 07 11:42:27 <[mbm]> I fint the "linux isn't posix complaint, let's remove pieces until it is" to be rather amusing Mar 07 11:43:01 <[mbm]> use a bsd system. it's not posix either, and a hell of a lot more annoying than the gnu stuff Mar 07 11:43:06 too bad we can't apply that to our government. Mar 07 11:44:54 anyways, off to sleep, going to the zoo tomorrow, maybe I'll give a router to the penguins. Mar 07 11:52:23 [mbm]: I usually just choose a different standard Mar 07 11:52:30 luckily there are many to choose from Mar 07 15:31:09 pavlov * r6535 /trunk/ (4 files in 2 dirs): split out scsi module support and add ata piix support Mar 07 15:38:11 pavlov * r6536 /trunk/package/kernel/modules/block.mk: forgot to set eol-style Mar 07 15:38:58 pavlov * r6537 /packages/utils/dialog/ (. Makefile): dialog for graphical prompting Mar 07 15:42:34 pavlov * r6538 /packages/lang/python24/ (. Makefile): cp python to python24 in prop to migrate to 2.5, if 2.5 succeeds completely, python24 can br rm'd.. otherwise ignore this for the moment Mar 07 22:24:30 hcg * r6539 /trunk/toolchain/gcc/Config.in: Changed default compiler for 2.6 kernels to 4.1.2 Mar 07 22:49:03 Kaloz: Fully working, well except crypto (the gateway) - see whether intel will give me the crypto code I guess **** ENDING LOGGING AT Thu Mar 08 02:59:57 2007