[IGSMAIL-4967]: Bug report and update of RINEX compression programs

[Yuki Hatanaka]

******************************************************************************
IGS Electronic Mail      12 Jul 04:29:24 PDT 2004      Message Number 4967
******************************************************************************

Author: Yuki Hatanaka

Dear colleagues,

This e-mail is to report a bug of the RINEX compress/uncompress 
programs (RNX2CRX and CRX2RNX) and to inform you the release of 
a bug-fixed version of the programs.

[Problem in ver 2.4.0 or earlier]
Resently, several users reported cases that numerical values 
of uncompressed RINEX files are different from those of the 
original ones. This problem appears when phase values are 
extreamly large. Exactly speaking, this happens when 
interger parts of phase (or code) values or their differences 
exceed a range of -2147483648~2147483647. This range comes from 
the use of long integer valuables which are usually 32 bit. 
I blindly assumed that the integer part of phase values are within 
this range if data are normal. According to the problem report by 
users, however, some of the new receivers (at least, Trimble NetRS) 
can output the phase values out of this range in some mode.
Although it is only integer part that can overflow, a phase 
jump caused by this bug is not always the same as integer 
cycle slip because the sign of integer part may be inconsistent 
with that of fractional part.

[Release of new version of the programs (ver.2.4.1)]
The software is updated to ver.2.4.1 with fixing the above bug.
It is available from following address:

    ftp://terras.gsi.go.jp/pub/software/RNXCMP_2.4.1/

The new programs were tested on LINUX, HP-UX, SUN Solaris, 
DEC (UNIX/LINUX), Windows 2000/98. The directory structure is 
slightly changed from the formerversion for convenience of users: 
Now *.tar (for UNIX) and *.exe (self extract exe for DOS) of 
binaries and shell/batch are available for the above OSs. 
There is no change in usage, and most of the users of the former 
version only need to replace the binaries. If you are using 
machines other than the above, the source code can be compiled 
by using ANSI C compiler. See docs/README file.

The modification other than the bug fix mentioned above are;
   - additional checking the resolution of double float valuable.
   - increase MAXSAT(-->100),MAXTYPE(-->20) 
   - additional checking for number of satellites and data types.
   - use of fgets instead of gets (crx2rnx.c)
   - fix a bug in splname.c (for DOS)
The running speed of the ver. 2.4.1 is 20-30% slower than ver 2.4.0, 
unfortunately.

I also ask IGS CB to replace the software with the newer one.
I apporogize for troubles you may have for the bug 
and recommend to update the software to the version 2.4.1.

Best Regards,
Yuki
-------------------------------------------------
HATANAKA,Yuki,  Geographical Survey Institute
                e-mail : hata at gsi-mc.go.jp
-------------------------------------------------

==== Contents of the directory ====

    docs/          the document of Compact RINEX format(crinex.txt)
                   and sample files.
    source/        C source code of the programs (common for UNIX, DOS, and VMS)
    unix/
         *.tgz     tar (gzip) package of binary with shells and docs/ for several OS
         shell/    front-end tools for multiple file handling (C-shell scripts)
    dos/
        RNXCMP.EXE self-exreact package for DOS-exe files with batch scripts and docs/
        exe_win/   DOS-exe files of the programs compiled under Windows2000
        batch/     front-end tools for multiple file handling (batch scripts
                   and a small C program)
    vms/
        com/       front-end tools for multiple file handling (VMS command
                   files)