Re: Error reading NML messages using Java

[Will Shackleford <shackle-at-cme.nist.gov>]

I suspect a strange version mismatch in the library. When I look at the 
source file rcslib/src/cms/cms_in.cc from the February2002 archive on source
forge you will see several lines  that look like this:

rcs_print_error ("CMS:(%s) Error reading from global memory at %s:%d\n",
		       BufferName, __FILE__, __LINE__);


however line 738 is this

  else if (queuing_header.tail < queuing_header.head)
    {
      free_space = queuing_header.head - queuing_header.tail;
    }

Either you  somehow got a copy of an earlier version of this file or 
the object compiled from an  earlier source.

------------------

Regardless there were some fixes that have to do with queuing  in
May 2002 and some fixes to the way Java handles arrays in NML messages in 
October 2002.  So it is probably worthwhile getting a later version either
the November 6 copy from the NIST ftp site or getting the source directly from
the sourceforge CVS repository.

The NIST ftp site is :

ftp://ftp.isd.mel.nist.gov/pub/rcslib/rcs_src.tar.gz

------------

You will probably eventually want to choose neutral=0 and "xdr". "xdr" is more efficient than "disp" , "disp" is useful if you need to directly connect via sockets with a programming language other than C++ or Java. You can set neutral=1 tosave space in the buffer when using dynamic length arrays  or to have hetorgenous processors directly accessing the same memory card on a VME backplane.



-- Will



-- 
---------------------------------------------------------------
William Penn Shackleford III			shackle-at-nist.gov
National Institute of Standards & Technology	Tel:	(301) 975-4286
100 Bureau Drive Stop 8230  			FAX:	(301) 990-9688
Gaithersburg MD  20899  USA
Office Location: Bldg. 220 Rm A253 
http://www.isd.mel.nist.gov/personnel/shackleford/