Hi Pete,

Why would a file be executable? In the tradition of the omni-
platform transportability of FITS files, it is unlikely that a file
would be really executable.

Just to provide some context, foreign file encapsulation was created
as part of a prior NOAO archive project several years ago to
encapsulate various non-FITS metadata products generated on Kitt Peak
– observing logs, weather pictures, etc. NOAO is using it again to
transport PNG files from the Mosaic camera pipeline to the next
generation NSA archive. It was designed to be very general purpose,
hence the authors' interest in reminding the community of its
existence. You can certainly create file collections that a third
party would be foolish to unpack on their own system. The same is
true of tar. In any event, if the fgread program isn't setuid, such
files would only be executable as the UID of user, containing any risk.

Nelson Zárate will have an ADASS poster, and perhaps we can expect
foreign encapsulation and the general topic of FITS extension types
to enliven the FITS BoF.

If the executable was intended to go onto a machine of the same
architecture (and libc....) then why use FITS?

We've long since moved past the point of regarding FITS as a pure
transport format. It's used in all our archives, for one thing. One
might well want to package up processing scripts with the data they
were applied to, for instance.

An example of an executable file, however, would be a virus.
Surely we don't want to go there?

One person's virus is another's intelligent agent. I suspect that a
lot of us have entertained the notion of an IRAF or IDL virus in the
past. Why not FITS? VO will certainly have its hands full with XML
and web services, many of whose features are intended to be very
"intelligent" indeed.

I suspect that foreign encapsulation is a sufficiently narrow niche
market that we need not fret too much.

Rob