Re: [ferret_users] what is ferret's limit to the full path name for a netCDF data set

[Ryo Furue <furue@xxxxxxxxxx>]

Hi Billy,

| sp ln -s /very/long/path/including/many/subdirectories/filename.cdf  
| temp_name.cdf
| use temp_name.cdf
| ..... work with the file .....
| sp rm temp_name.cdf   ! clean-up after use
| 
| Don't forget the -s option! (symbolic link). Otherwise rm will
| remove the actual file!

Sorry for this nit-picking, but rm won't remove the actual file when
it deletes a new hard link.  To simplify, a filename is a hard link to
data on Unix.  When we talk about a "file", we mean the data and the
filename pointing to it.  When we create a hard link as

   $ ln oldname newname

both "oldname" and "newname" point to the same data.  Now, if we
delete either oldname or newname, the data still remains because there
still exists a name pointing to it.  The data will be thrown away only
when no name is pointing to it (*):

   $ rm newname  # -> oldname still points to the data
   $ rm oldname  # -> no name points to the data.

Regards,
Ryo
--------
(*) It's a kind of reference-counted garbage collection, for those
   of you with CS slant.