20001002: SHEF decoding sparse

>From: "Arthur A. Person" <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200010021930.e92JUab10348

>Hi...
>
>We've noticed a problem lately that a lot of SHEF data is not being
>decoded by dcshef.  The logs have a lot of messages like:
>
>DCSHEF[23323]: 001002/1927: [DC -7]  The buffer has overflowed, no end of
>bulletin.
>
>and if I run it manually using the raw SHEF data stream from a file, I get
>the following same kind of error:
>
>DCSHEF[12208]: 001002/1904: Starting up.
>DCSHEF[12208]: 001002/1904: read 10240/16383 bytes strt 0 newstrt 10240
>DCSHEF[12208]: 001002/1904: Invalid date/time: SRUS53  KMQT    010000
>DCSHEF[12208]: 001002/1904: Times do not match: 001001/0000    001002/1904
>-2584
>DCSHEF[12208]: 001002/1904: read 6144/6144 bytes strt 10240 newstrt 16384
>DCSHEF[12208]: 001002/1904: read 553/553 bytes strt 0 newstrt 553
>DCSHEF[12208]: 001002/1904: [DC -7]  The buffer has overflowed, no end of
>bulletin.
>DCSHEF[12208]: 001002/1904: Normal Termination.
>DCSHEF[12208]: 001002/1904: Number of bulletins read and processed: 1
>DCSHEF[12208]: 001002/1904: Shutting Down.
>
>Any ideas?
>
>                                Thanks.
>
>                                  Art.
>
>Arthur A. Person
>Research Assistant, System Administrator
>Penn State Department of Meteorology
>email:  address@hidden, phone:  814-863-1563
>


Art,

I looked in my Oct 1, 00Z file (trying to match up with the above data you 
have),
I found a bulletin SRUS56 KLOX that exceeds the 16384 character buffer used by 
the 
dcgbul routine to get a single bulletin. This is the buffer that is overflowing.

There are 2 ways to fix this. 1, recompile GEMPAK with DCMXBF increased in
bridge.h and bridge.prm. The program itself alread uses DCMXBF*2.
Alternatively, the program could attempt to re-enter the read loop
instead of exiting.

The quickest fix would be to limit the pattern of files going to the decoder
to eliminate the KLOX product.

Steve Chiswell