Detecting mount containing running script

[schragge]

No. Why should that be a problem? It's syntactically correct. The else branch just won't be reached in this case. Open a terminal and try something like:

(if true;then exit;else some garbage;fi)

[kiplingw]

No. Why should that be a problem? It's syntactically correct. The else branch just won't be reached in this case. Open a terminal and try something like:

(if true;then exit;else some garbage;fi)

Right, but how will you know if the "exec eject" command completed successfully? As soon as it's run, execution leaves the script and never returns again, so there is no way to check eject's exit code and to do something based on that.

[ofnuts]

It would be nice if the script were not locking the device and control could return to the script after the eject command. But I guess that is not so straightforward.

It's not the script that is locking the device...

A different approach is to have your script self-copy to /var/tmp and then 'exec' the copy (maybe passing its original location as a parameter or environment variable). Then you are no longer using a file on the device.

[schragge]

+1 to what ofnuts said.
There may be a problem if /var/tmp (or /tmp if you choose to use it instead) is mounted noexec. In this case, invoke it like

exec bash /var/tmp/$scriptname

[kiplingw]

+1 to what ofnuts said.
There may be a problem if /var/tmp (or /tmp if you choose to use it instead) is mounted noexec. In this case, invoke it like

exec bash /var/tmp/$scriptname

It's definitely possible to copy itself to tmp and execute the copy with the idea that it would relieve the device it was hosted on from being locked. I had considered that route, but then I figured it might not be worth the trouble and probably easier to just find a way to run the exec eject as the last command. But thanks for everyone's help.