webconverger / webc Goto Github PK

so if homepage contains "debug" it shouldn't trigged debug mode

I tried to regenerate the initrd inside the chroot (for installing a different kernel), but that resulted in a non-bootable system. The same happens when I just regenerate the initrd without changing the kernel.

To reproduce (inside chroot):
# aptitude reinstall initramfs-tools live-boot
# update-initrafms -u -t

Then build an image and boot it. Inside virtualbox, I get a panic "attempted to kill init", so I presume some required customization was lost, but I can't really find what customizations were applied?

https://groups.google.com/forum/#!topic/webc-users/7VvFvf8yq2k/discussion

It would be good to have printing support back as people ask for it from time to time.

I've been holding out for gitfs to be mainlined, which should radically reduce the size of the ISO, making it reasonable to add the 100M or so CUPs support to Webconverger.

Ideally Patrick will mainline the gitfs work he has been doing: https://github.com/Webconverger/webc/tree/git-fs-setup soon!

http://www.youtube.com/watch?v=xBEbpDTeWPE&feature=youtu.be&a

I tried reproducing this in debug mode at first and just a normal build under Virtualbox. Sound works fine. I'm hoping he can provide more details of his setup.

Hi there,
could be that Im doing soemthing completely wrong, but I tried to add the cronjob you have posted, but it does not work.
I have selected my timezone and added the line you have posted to my boot options. At the given time just nothing happens :).

When I go to debug this it gives me out the right time, but syslog gets spammed with /etc/webc/updates.sh messages, and as far as I could see there was nothing related to cron.

I would like to use this in the live version.

https://groups.google.com/d/msg/webc-users/NZS6C1dIqSg/DzVfqQPKiwgJ

Just tried http://webconverger.org/blog/entry/USB_tethering/ with ICS Android and it didn't work until I manually run dhclient usb0 during debug mode.

Thanks to Matthijs Kooijman for rightly raising this

We need some code to switch the libGL.so.* symlinks between nvidia and libgl1-mesa-glx
https://github.com/Webconverger/webc/blob/master/etc/webc/startx.sh
http://anonscm.debian.org/viewvc/pkg-nvidia/packages/nvidia-graphics-drivers/trunk/debian/

I notice that you get the UUID from the mac of "eth0". Is this safe?

1. Does udev these days ensure machines with multiple network cards never switch interface names?
2. Don't some machines boot with interfaces not named "ethX"?
3. Is the interface that's brought up also harcoded to eth0?

https://github.com/Webconverger/webc/blob/master/etc/network/interfaces currently gets wiped out by a d-l script in the initrd

This is related to #13

We rely on some of the scripts from https://github.com/Webconverger/webc/tree/master/lib/live/config to run for:

• timezone setup (this could be moved somewhere else... but where?)
• setting ssh keys (this may be preset and not needed)

From https://github.com/Webconverger/webc/blob/master/lib/live/config.sh#L210 we can see it needs live=config not live=local

Add pxe bootable images to our downloads

I think this is an issue with how https://github.com/Webconverger/webc/tree/master/etc/network/if-up.d is run in Debian

https://groups.google.com/d/topic/webc-users/Wx6GWJxcJLU/discussion

It should return the mac it's actually using

Or if config.webconverger.com is down for an extended period of time.

Currently the status quo is that it will reset to defaults after 15 seconds of trying.

We need to figure out and define this user experience.

Can't we do away with these legacy Debian live configs?

When I try remove those /proc/cmdline options I get X1 respawning too fast. So atm we still depend on something in the Debian Live initrd IIUC.

I tried to install Webconverger 12.3 and the daily build to the hard drive but after it restarts, it won’t boot. I just get a blinking cursor in the top left corner. I tried installing from the CD and USB using both builds. I have tried this on a Dell Latitude D610 and VMware, just to make sure it was not a hardware issues. I can boot the Live CD/USB just fine and everything works, it just looks like the HDD
boot loader it not being loaded correctly.

See:

http://farm8.staticflickr.com/7121/7042291359_e69d9fd541_z.jpg

That is with:

webc-12.2-16-g6f05d61.iso 03-Apr-2012 11:51 585M

When machine is shutdown, the root partition should try and shutdown safely.

Perhaps there is some short cuts that can be made here.

How about changing the /etc/webc/configuring.png icon to that found here: http://www.iconfinder.com/icondetails/8844/128/kit_tools_wrench_icon. The configuring.png icon does not have a transparent background, but no-net.png does.

Since no-net.png is from the same Oxygen GPL icon set, and the look is very similar and http://cdn1.iconfinder.com/data/icons/oxygen/128x128/categories/preferences-system.png does have a transparent background, it make sense to use it.

I know, it's such a small picky thing. This would allow a white or black boot screen during startup though.

Suggestion from Daniel:
the usual compromise would be to offer 686 by default and a 'fallback'
with 486, so that the majority of the users is not hurt by the 'special'
needs of the minority with very old systems (we're talking here about
machines from before 1997 where pentium2 was introduced, and, of new
'low-end/power-saving' appliance like machines of today).

For http://webconverger.org/filtering/

Could this happen at the first load after an upgade? That seems to be what happens.

It seems pointless to show a single tab when most people are only ever going to have one tab and cannot create a new tab. The setting is browser.tabs.autoHide = true.

http://build.webconverger.org/webc-498e493.iso

Booting on my work's network with two different machines, both show the network disconnected screen for exactly(?) 30 seconds before loading firefox and loading homepage successfully.

For statistics

https://github.com/Webconverger/webc/blob/master/etc/rc.local#L6 must be the culprit

Indeed it's because eth0 doesn't exist on boot. Somehow udev names it eth1 and eth0 is some other interface...

http://flic.kr/p/by4uJ9

How about something like for i in eth{0,1}; do ip address show dev $i 2>/dev/null | awk '/ether/ { print $2 }'; done

or would the better fix happen with udev?

That being said, that script has several issues:

• it uses gpg --verify's exit code. That's horrible. Use gpgv or gpg's
  --status-fd and look for GOODSIG or VALIDSIG.
• you accept every key in the keyring for signing this thing. no matter
  who signed it. That's also not good.
• Which key is authorized to sign this thing anyway. What is this
  thing? Can it not ship its own key updates?
• Who runs this thing? How many keys will they have in their public
  keyring? Why do you think you should update them all?
• After you fetch keys, you probably do a trust-db update. That'll also
  take ages if there are a lot of keys in the keyring. And you probably
  shouldn't care about who signed what or who is trusted and how many
  marginals are needed etc anyway.

Maybe: you want to use gpgv, with a trusted-keyring.gpg, that gets
updated by some other means. That'll take care of most of that.

http://lists.x.org/archives/xorg/2012-April/054494.html

https://github.com/Webconverger/webc/blob/master/usr/bin/kioskresetstation

This should give better compatibility[1] and performance.
[1]https://groups.google.com/d/topic/webc-users/0lTYXqQLkJw/discussion

http://us.download.nvidia.com/XFree86/Linux-x86/290.10/NVIDIA-Linux-x86-290.10.run
Doesn't work as it refuses to install since it can't detect nvidia GPU on builder.

http://webconverger.org/security/
80bba79

https://github.com/Webconverger/webc/blob/master/home/webc/.xinitrc#L56
-layout "us, th" -option "grp:shift_toggle"

I think it needs to parsed in to a array and rebuilt http://mywiki.wooledge.org/BashFAQ/050

2f2667e

Something to watch out for.

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=668264

If we can reproduce it, then we can simply checkout an older version in the interim.

http://www.flickr.com/photos/hendry/6880067186/

This is potentially quite serious, since SSL will fail typically with bad date/time

With an Asus UX31E with USB ethernet:

Worked fine with 2012-03-21.iso: I booted and subscribed.

With 2012-03-25.iso I am greeted with the network disconnected graphic.

Booting with "debug" I can see that the hardware is still recognised (ifconfig -a shows the interface) but I don't know how to attempt bringing up the network manually.

The ethernet connection works fine booting from my normal Arch install.

Re-test timezone= and introduce a halt= crontab mapping API

http://webconverger.org/cronjob/ <-- needs updating

With a previously subscribed X220 I am greeted with the webc unsubscribe page on boot, instead of my chosen home page and chrome.

Only /var and /tmp are write-able, so symlink the $HOME/.mozilla to
/tmp/webc-mozilla then mount them as no-exec.

live_config_update isn't safe

http://webconverger.org/startup