Comments for OPERA:

1) File desc.c: On Line 100: the read_table_line() is commented out.
This causes compiler warnings because of an unreferenced function.  
You should remove the function entirely, if not needed.

2) File decbufr.c: The call to get_lens() is still being passed the total
length of the file in "len", even though the GTS header may have been
stripped off and the starting pointer "buf" has been advanced.  The
checking will thus allow reading beyond the end of the file.

3) File decbufr.c:   i = fprintf()  gives warning messages due to
unreferenced variable.

4) Files desc.c, bufr.c:  Stray printf's cause data in standard out.  
I am using the code in a pipe, so I changed all these to fprintf's
to stderr.  It might be nice to have an option to be silent.


Comments for Rosa Dias:

Need to fill in all values in config structure.

All times are converted to GMT, even if not orignally?

I assume in 0 04 001 year is full year 2002, etc.

I assume in 0 04 002 month is origin 1, range 1-12.

Since order should not matter, you should add a new code to BUFR
for 4 corner lat/lons.

=============NEW COMMENTS=====================

BUFR product types has no SRI or SRT (maybe use CAPPI?)
Also has type composite.  How do we flag CAPPI composite?
Does composite take precidence?

Page 19 says that 4=Gnomonic projection.  Manual on Codes says 0=Gnomonic.

I assume that scale factor can be incremental at the center
or at the reference point.  Also that the false easting, and 
northing are using these scale factors.


GTS product types has a very different list from BUFR.  For
example, where is PPI?  I assume horizontal resolution is
OK.  What do we put in for invalid product types?  I will put
in a "Z".

LBUFLEN needs to be brought out into a header file so a program
can test to make sure things fit OK.



