NOAA / NWS's Meteorological Development Laboratory
| |||
Degrib (NDFD GRIB2 Decoder) |
NDFD Home | Verification | News | Organization |
|
Subject: Degrib 1.31
Date: Fri, 24 Oct 2003 12:04:24 -0400 From: Arthur Taylor To: arthur.taylor@noaa.gov Hi You all have the unique distinction of being registered on my degrib (aka NDFD_GRIB2Decoder) mailing list (all 214 of us) (I'd like to welcome: Shawn Truelson, Mark Gavin, Tanja Fransen, takmaki, deanchurchill, Daniel Chace, alpha2moon, douglas, scsebnic, and A. P. Benney) (A special hello to Tanja Fransen, who I'd like to thank again for her hospitality.) Please let me know (email me) if you want to unregister. The latest version of the program is on: www.nws.noaa.gov/mdl/NDFD_GRIB2Decoder/ ----------- General News. Correction: In my release notes for degrib "1.30", I mentioned a NWS focus article where the editor was interested in NDFD Customers... I had a typo and said that it was 9/22/2003. It was actually 9/23/2003: http://weather.gov/com/nwsfocus/fs092303.htm ----------- News about this release: I have just finished putting together version "1.31" (last one that I announced was "1.30" on 10/20/2003). This release is basically a bug fix, combined with a slightly improved "get data" routine. 1) Bug Fix: For "the big polygon". In 1.30, this worked fine in MS-Windows, but as Matthew Kallio pointed out, it performed a "segmentation fault" in linux. Matt was kind enough to email the solution, which turned out to be a one line addressing error when using fwrite(). I'm still not sure why it worked at all in MS-Windows, but the corrected code should be safer for MS-Windows, and get it to work in unix. 2) Tanja reminded me again, that there is an instant on the server where the NDFD data files are not complete. This may be because of copying it from one place to another on the server, or it may be some other issue, but the result is that it is possible to download an incomplete file, which causes degrib problems. Solution: I have changed the download method in degrib1.31. It now takes advantage of the fact that the first part of the file specifies the total file size. tkdegrib compares that number to the actual file size. If they don't match (incomplete file), it puts it in a list of files to try to re-get. After it finishes getting all the regular files then it re-tries to get the files in its "incomplete file list". This should stop the "incomplete NDFD file" issue from affecting tkdegrib, since by the time it tries to re-get the file, the file should be there. The only exception I can see, is if it was the last file in the set, but even then, by the time degrib downloads the incomplete file, determines it's incomplete, and starts to re-get the file, one would "expect" the file to be there. This is a very difficult bug to test against since the files are only incomplete for an instant or two every hour, so even though I think tkdegrib is "safe" now, I could have missed something. ----------- Please let me know if you have problems. Regards, Arthur (Art in MDT (Mountain Daylight Time) :-) ) ("Big A" for Glasgow's WCM :-) )
|