On Wed, May 22, 2002 at 04:59:37PM +0200, Lars Weber wrote: > Is this also true for passive translators? Do they also not store the > path to the translator executable (as I've thought until now) but a direct > reference to the file instead? If so, what would happen if the translator > is replaced by a newer version for example?
They store the complete filename, but I am not sure right now about their execution context. I would expect as they are started by the parent filesystem, they get to see whatever the parent filesystem sees at this filename. > Or is the path for passive translators stored in the inode but looking up > the executable does not cause the same translator to be queried again? When the passive translator is started, the will be an active translator on the node. Once there is an active translator, the passive translator will not be restarted. > Anyway, I'm just curious. If answering my questions above would take to > much effort I'm as happy to be told that there is no problem and I needn't > worry... > > Or maybe the information I need is available somewhere in the docs? In my talk at http://www.gnu.org/software/hurd/hurd-talk.html I am covering the exact mechanism by that filenames are looked up and translators are started to some detail. However, I don't cover the complete execution context of passive and active translators, or even any particularly weird example. There is one email on the list from somewhere in 2001 where Neal explains the difference of the execution context of passive and active translators, if you are really that interested, it would be easy to dig that out, contact me privately. Thanks, Marcus -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]