* Russ Cox <[EMAIL PROTECTED]> wrote: > rfork n forks the name space, so that any changes > made by the script don't propagate out to the > parent process. This means that (1) vacfs can > mount on /n/vac without any fear of bothering > some other instance of the script, and (2) when the > script exits, there will be no more references to > the mounted /n/vac and the kernel will unmount it.
Aha. But this still does not terminate the vacfs, right ? > > Meanwhile I've hacked up vacfs to give it an "autodie" mode. > > It means: automatically terminate if the last file has been > > closed. Seems to work, but now idea how stable it would be > > in production ;-o > > In the time it took you to do that you could probably > have written unvac instead. hmm, maybe ;-o Meanwhile I've written an new tool which stores single files (no directory stuff at all). Not perfect yet, but already works quite fine (IMHO): svn://nibiru.metux.de/public/plan9port/apps/vtstore/ In the next step, I'll add http-alike metadata (mimetype, etc). cu -- ---------------------------------------------------------------------- Enrico Weigelt, metux IT service -- http://www.metux.de/ cellphone: +49 174 7066481 email: [EMAIL PROTECTED] skype: nekrad666 ---------------------------------------------------------------------- Embedded-Linux / Portierung / Opensource-QM / Verteilte Systeme ----------------------------------------------------------------------
