[Pythonmac-SIG] macostools.copy failing after volumes are mounted or unmounted during script execution

[Paul C. Pharr]

Hi everyeone,

I have attempted to use the macostools.copy call under python 2.3 (stock install under Panther) to copy a large quantity
of files to a file server from a python script which runs continuously as a server process. After recently changing the
script to dynamically mount the server volumes as necessary using the command line "mount" command, I started getting
seemingly random, but frequent failures referencing a -120 Directory Not Found error. After investigation, it turned out
that this failure always happened on the third copy operation to a new directory.

Thinking I had found a subtle MacOS X bug, I submitted a request to Apple's Developer Technical Support, and after a few
days got a long and informative response which said essentially that:

The Carbon File Manager, opon which macostools is based, depends on being notified within the process's runloop when
volumes are mounted or unmounted within the file system. When I make a system call from my script to mount or unmount a
volume, my script and its instance of the Carbon File Manager is not notified because it does not have a runloop.


My question to you Mac Python experts out there is this:

I have seen some references to Mac python scripts implementing a runloop to respond to system needs. Is there a
standard, simple (hopefully trivial) way to do this which would allow Carbon File Manager to get the notification it
needs? Is there a complicated way to accomplish this?


Alternately, does anyone know of smething I can do within my script after mounting or unmounting a volume from within my
script which will give CFM a kick so it will requery the state of all mounted volumes and eliminate the need to respond
to these notifications?

Many thanks in advance,


Paul Pharr