martin f krafft <[EMAIL PROTECTED]> writes: > also sprach Martin Dickopp <[EMAIL PROTECTED]> [2004.10.26.2128 +0200]: >> When the device is pulled from the USB port, the script kills all >> processes that have been accessing the filesystem and then >> unmounts it. > > Have you tried the -l option to umount instead? > > Imagine the case where I am working on my laptop on a file sitting > on a stick, and someone in the train wants me to get up while I am > adding the finishing touches to a paper. As I get up, the stick > slides out a little, causing my vim to be killed and potentially > taking all data with it?
Good point. > Why kill? The idea was that once the device is pulled, everything is lost anyway. I have to admit that I had failed to think of the case where the same process also holds data not stored on the device. I'll kill the killing of processes. :) > USB handles insertions correctly, giving power to the device first > before connecting the control. It's crap that they didn't go one > step further. Indeed. >> I must have misunderstood your point about using floppies instead, > > The speed. Using sync means that you are going to crawl. Given that the user can pull the device at any time (even accidentally, as in your example), I think that data integrity should take precedence over speed. Anyway, if most other users use their USB pens in the same way as I do (i.e. primarily for data exchange, not as a working directory), speed is not such an important issue. Martin -- ,--. ,= ,-_-. =. / ,- ) Martin Dickopp, Dresden, Germany ((_/)o o(\_)) \ `-' http://www.zero-based.org/ `-'(. .)`-' `-. \_/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
