[iaufwg] approval of the 'FOREIGN' extension convention

[Mark Calabretta]

On Wed 2006/12/06 15:44:38 CDT, William Pence wrote
in a message to: IAU-FWG <iaufwg at nrao.edu>

>Please review this document for yourself, and let me know if you have
>any further comments or objections about this convention;  If no one has
>any objections, then the FOREIGN file encapsulation convention will be

I assume you mean comments or objections about the documentation of
this convention, not the convention itself.

The first sentence of Sect. 2 would be better as

  A FITS extension of type 'FOREIGN' (henceforth a "FOREIGN file
  extension" or just "FOREIGN extension") provides a mechanism for
  storing an arbitrary file or tree of files in FITS, allowing it to
  be restored to disk at a later time.

It is not actually stated anywhere that a FOREIGN extension stores a
single file.  (Or if that is not true then there's even more explaining
to do.)

It is not explained how "The FG keywords are used ... in standard FITS
extensions such as IMAGE, BINTABLE, and so on".

In section 3, the keyword name should be followed immediately by the
keyvalue type, e.g.

   FG_GROUP (string) - Each time a ...
     :
   FG_LEVEL (integer) - The directory ...

I would have though that implementing the FG_TYPEs of "directory" and
"symlink" would be a simple matter of naming them and, for symlinks,
the target.  Are they really "implementation dependent"?  If so, this
requires more explanation.

Why is it necessary to distinguish an FG_TYPE of "FITS", and moreso
"FITS-MEF"?  Couldn't any general FITS file be stored as type "binary"?
It seems that FITS files are treated specially but without explanation.

With respect to FG_LEVEL, it is not adequately explained how a directory
tree may be reconstructed, i.e. how is a file associated with a
(sub-)directory?  I don't see how it can be done.

What is the distinction between FG_SIZE and PCOUNT?

FG_FMODE is not adequately explained - it clearly shows a unix bias.
How would it be interpreted by, say, a Windows programmer who knew
nothing about unix?

The three sentences beginning "When a file group is restored ..."
make no sense to me.  Should "foreign" here in lower case be upper
case?

Much of Sect. 4 seems irrelevant to me.  How would it help someone to
create or interpret a FOREIGN file extension?  Instead, there should be
more explanation of how the basic types "text", "binary", "directory"
and "symlink" are meant to be stored, i.e. not as an implementation
issue.

Typos noted in passing:

"with following five" -> "with the following five"
"adecade or two" -> "a decade or two"


Cheers, Mark