Re: qsh find command -size +20M

[Scott Klement <midrange-l@xxxxxxxxxxxxxxxx>]

Rob,

The "FTP must be binary" is your own fault :)  Or perhaps, is a 
byproduct of the fact that you started this in QShell and later switched 
to PASE.

Qshell understands CCSIDs and character translation, as it is native
IBM i stuff and therefore understands IBM i CCSID support.  When you 
output from QShell, the file will be created as CCSID 37 (EBCDIC).  It's 
a native IBM i language, so EBCDIC is it's default.  If you had 
pre-created the file in ASCII (which you didn't... but hypothetically..) 
 QShell would understand that, and would automatically translate to 
ASCII (or whatever the file is)

PASE is not like that.  PASE (QP2TERM/QP2SHELL) is not native IBM i 
stuff, it's Unix (AIX) stuff.  It doesn't understand CCSIDs on files, it 
only understands ASCII, that's it.  So when you run your code in PASE, 
it outputs ASCII.  Period.  Your file already existed (from the earlier 
QShell examples) and was marked as EBCDIC -- but PASE didn't care, it 
put ASCII in it anyway, because that's all PASE understands.

Thus the problem.  You now have a file that's announcing itself as 
EBCDIC ("Hi, I'm an EBCDIC file") but it contains ASCII data.  When you 
FTP that in ASCII mode, it'll attempt to translate it from EBCDIC to 
ASCII (and since it's not actually EBCDIC, you'll have a mess)

Solution is simple.  Delete the file, then do it purely from PASE.  The 
file will be marked with CCSID 819, and therefore translation will work 
correctly.

Or, use QShell instead of PASE.  if adding the "M" for megabytes is the 
only reason you're using PASE, then I'd say the suffix isn't THAT 
important.  Think about it.  How hard is it actually to take 20,000,000 
and divide it by 512?



rob@xxxxxxxxx wrote:
> Well I pretty much got this to work now
> find . -type f -size +20M -exec ls -l {} \; | awk '{ print $9 "," $5 "," 
> $6"/"$7"/"$8 }' > /rob/bigfiles2.csv
>
> Had to use binary to ftp that down.  Ascii looked like gibberish 
> (CCSID=37).