severity 743360 minor thanks OK, I've found the problem and it was not in fuse but in a test package locally.
It would be very useful if the fuse postinst was fixed though: if [ -x /sbin/MAKEDEV ] && [ ! -e /dev/fuse ] then echo "Creating fuse device..." cd /dev MAKEDEV fuse chmod 0666 /dev/fuse else echo "MAKEDEV not installed, skipping device node creation." fi This is misleading. The error message conflates the lack of MAKEDEV with a missing /dev/fuse which means that if /dev/fuse exists but makedev is not installed (and udev is the primary alternative in the fuse dependencies), then fuse reports a misleading error. fuse then goes on to try udevadm control --reload-rules without checking that the fuse module has been loaded, which results in udevadm outputting "device not found". Together with the message that the device node creation has been skipped, this leads to predictable confusion. The device existence should be checked separately and the "skipping device node creation" message omitted *only* if MAKEDEV does not exist and /dev/fuse does not exist. Currently, the else applies to either condition. Then the postinst should check the output of lsmod and produce a helpful message about the fuse module preferably *without* failing to install the package. So there are three bugs here: 0: fuse fails to install if the module has not been loaded - confirmed & reproduced 1: fuse blames the lack of a device node when it is the lack of the module which is the problem, yet fuse does not even check lsmod 2: fuse reports that the device node was not created because MAKEDEV is not installed when MAKEDEV has nothing to do with the actual problem - the device node exists, it just has no kernel module to go with it. Please fix the postinst to be much more useful and to correctly handle the lack of the fuse kernel module *without* causing the package installation to fail but *with* a noisy warning on package installation / configuration. -- Neil Williams ============= http://www.linux.codehelp.co.uk/
signature.asc
Description: PGP signature