[daip] A question about migrating old AIPS data files

[Eric Greisen]

Tony Willis wrote:
> Hello Eric
> 
> Recently I've been collecting some data with the DRAO synthesis
> telescope which I'd like to compare with very old WSRT data taken
> about 25 years ago. I'm crossing my fingers that the WSRT data is on
> an old exabyte tape (backed up in 1992) which I managed to read 
> successfully.

We have always strongly recommended the use of FITS for backup which 
would avoid all the problems discussed below.

> 
> Of course a cursory examination of the files (a partial listing is
> attached) reveals that if nothing else, when compared to a modern
> AIPS directory listing, you now use 9 digits to the left of the
> decimal point vs 8 in the old data files. However I'm sure the
> differences run deeper than that - I found a python script on the net
> that will read and print the contents of a medern AIPS catalog but it
> just threw up when I tried to read the CAC00000.08F;1 file! I'm also
> uncertain about the byte order of the data - I think that the original
> processing was done on a Sun workstation, which if I remember correctly
> was big-endian, whereas now I'm using a linux PC.
> 
> Anyway I wonder if there's any AIPS task or script which can migrate 
> these old
> files into a modern format? If not, can you steer me toward some docs
> that give the current file formats? I found the old 'Going AIPS' docs on
> the web - since they claim to have been last updated in 1990, they 
> presumably
> describe the old format.

The stand-alone program UPDAT does the file renaming from base 16 to 
base 36 with more characters.  Other format changes should occur 
transparently (in the D version we have managed to do translations 
internally on the fly) .  The problem as you surmise is the byte order. 
  You will need a big-endian version of AIPS to run UPDAT.  There is 
then a program (stand-alone) REBYTE which does the byte translation.

I suspect that - if you are decent at scripts or can find someone who is 
- that UPDAT almost completely is just a rename function.  Change

     XXCnnnmm.uuu;1

to
     XXDNNNMMM.UUU;

where nnn, mmm, uuu are in base 16 and NNN, MMM, and UUU are the 
corresponding numbers in base 36 (0-9,A-Z).  Then a modern REBYTE 
program can fix the data.  See HELP REBYTE inside aips.

Eric Greisen


> ================
> drwxr-xr-x 14 twillis twillis     4096 2009-09-08 15:48 ../
> drwxrwxr-x  2 twillis twillis     4096 1992-09-10 14:33 ./
> -rw-rw-r--  1 twillis twillis    13312 1992-09-10 14:33 CAC00000.08F;1
> -rw-rw-r--  1 twillis twillis     2048 1992-09-10 14:14 CBC01801.08F;1
> -rw-rw-r--  1 twillis twillis     8192 1992-09-10 14:14 HIC01801.08F;1
> -rw-rw-r--  1 twillis twillis  4194304 1992-09-10 14:14 MAC01801.08F;1
> -rw-rw-r--  1 twillis twillis    14336 1992-09-10 14:09 CCC01801.08F;1
> -rw-rw-r--  1 twillis twillis     2048 1992-09-10 13:50 CBC01501.08F;1
> -rw-rw-r--  1 twillis twillis     8192 1992-09-10 13:50 HIC01501.08F;1
> -rw-rw-r--  1 twillis twillis  4194304 1992-09-10 13:50 MAC01501.08F;1
> -rw-rw-r--  1 twillis twillis    14336 1992-09-10 13:46 CCC01501.08F;1
> -rw-rw-r--  1 twillis twillis     2048 1992-09-10 13:25 CBC01301.08F;1
> -rw-rw-r--  1 twillis twillis     8192 1992-09-10 13:25 HIC01301.08F;1
> -rw-rw-r--  1 twillis twillis  4194304 1992-09-10 13:25 MAC01301.08F;1
> -rw-rw-r--  1 twillis twillis    14336 1992-09-10 13:21 CCC01301.08F;1
> -rw-rw-r--  1 twillis twillis     2048 1992-09-10 12:55 CBC00F01.08F;1
> -rw-rw-r--  1 twillis twillis    10240 1992-09-10 12:55 ANC00F01.08F;1
> -rw-rw-r--  1 twillis twillis     2048 1992-09-10 12:50 CBC00E01.08F;1
> -rw-rw-r--  1 twillis twillis     8192 1992-09-10 12:50 FQC00E01.08F;1
> -rw-rw-r--  1 twillis twillis     3072 1992-09-10 12:50 HIC00E01.08F;1
> -rw-rw-r--  1 twillis twillis  4008960 1992-09-10 12:50 UVC00E01.08F;1
> 
> 
>