Wrong values in .data numpy array

[mraspaud]

From cpau...@gmail.com on February 18, 2013 16:25:48

What steps will reproduce the problem? 1.Reading Attached BUFR file with python-bufr
2.looking at the .data attribute of e.g. YEAR or LATITUDE (HIGH RESOLUTION) What is the expected output? What do you see instead? a numpy array filled with values but I only see the first value followed by 0. or the BUFR fill value What version of the product are you using? On what operating system? I am using python-bufr 0.2-5 in python 2.7 on Linux Mint 14 - 64 bit Please provide any additional information below. I'm trying to read ASCAT data with the python-bufr library. I checked the file with the ECMWF BUFR checker http://www.ecmwf.int/products/data/d/check/ and am also able to read it with our other BUFR readers.

I am no expert on any of this so maybe I compiled something wrong but I tried several times and it works but I can't read the data.

Best Regards,
Christoph

Attachment: W_XX-EUMETSAT-Darmstadt,SOUNDING+SATELLITE,METOPA+ASCAT_C_EUMP_20130201000001_32623_eps_o_125_ssm_l2.bin

Original issue: http://code.google.com/p/python-bufr/issues/detail?id=9

[mraspaud]

From cpau...@gmail.com on February 19, 2013 08:42:25

The same thing happens when I compile it in cygwin on windows 7. Could this be python 2.7 related or does it only work on ubuntu 10.04?

Best Regards,
Christoph

[mraspaud]

From marjolai...@gmail.com on May 24, 2013 02:27:50

Has this issue been resolved?
I am having the same problem. I am running python 2.7.3 with python-bufr 0.2-5 installed on a 64bit machine (Ubuntu 12.10).

[mraspaud]

From cpau...@gmail.com on May 28, 2013 03:41:15

The issue has not been resolved for me when compiling from source but I found a workaround:

Install the .deb package as explained here: https://code.google.com/p/python-bufr/wiki/InstallationUbuntu It works even though it is for an older ubuntu version. I tested it on Linux Mint 14 and Ubuntu 12.04.

I'm currently using it with python 2.7 even though the package was compiled for python 2.6.

[mraspaud]

From rnbokho...@gmail.com on November 19, 2014 05:20:14

The bug occurs for BUFR messages with multiple subsets. I digged into the code (_BUFRFile.c) and fixed it. Find the fixed source file attached and a patch file.

To install:

1. download python-bufr-0.2-5.tar.gz (from Downloads)
2. untar and cd python-bufr-0.2-5
3. replace bufr/_BUFRFile.c with the attched one
4. python setup.py install

It hasn't been fully tested but it works for my BUFR's...

Attachment: _BUFRFile.c-0.2-6 patch.txt

[mraspaud]

From cpau...@gmail.com on November 19, 2014 05:38:28

Wow Thank you. I will try this fix as soon as possible.

Unfortunately it seems that the development of this project is kind of stuck.

[mraspaud]

From rnbokho...@gmail.com on November 19, 2014 07:58:34

Yes so it seems..unfortunately.
However, this Python module is just a thin wrapper around the ECMWF BUFR decoder which is very alive. That's why I found it worthwhile to put energy in this :-)