Shell Programming and Scripting

Hi,
I have two large files with uid's:
- 581004 File1.txt
- 292675 File2.txt

I want to know which uid's are in File1.txt and not in File2.txt.
I have used comm -23 File1.txt File2.txt. This should do the trick i thought. But in the output i keep having uid's in File1.txt that are also in File2.txt, so not unique!!
I got rid of all the spaces, sorted them etc. (sort -u).
Has anyone any other options i can try?
Are my files too big?

I have tried the comm statement with smaller files with just a few names and than it works the way it should.
Tine

Try...

grep -vf File2.txt File1.txt

... but comm should work, perhaps its limit has been exceeded.

What is this:
- 581004 File1.txt
- 292675 File2.txt
supposed to mean? Number of lines? Characters? Or something else?

You give two theories as to the contents of the files: uids, and names. What kind of data do you really have?

How did you get rid of the spaces? What sort parameters did you use? Really just sort -u?

My guess is that you don't have the files properly sorted for comm, but I can't tell for sure with the sparse information you provided. Post some of the input that you're feeding into comm.

The numbers are the size of the file..

The content of the files i want to get the unique values are uid's, the files i used to test with contain names:
I got rid of the spaces by using sed 's/ //g' File.txt
To sort the file i used sort -u File.txt

These are a few of the uid's i have in the files:

a01
ali01
aloy01
araby01
atjak01
aa01
aa02
aa03
aa04
aa05
aa06
aa07
aabouz01
aachic01
aaddab02
aaddi01
aadrou01
aafar01

some extra info: File1 contains 3500 uid's, File2 contains 6500 uid's

Those files don't seem very big. I tried some tests with comm and I could not get a failure. I suppose that you could have buggy version of comm. Can you try it on another system?

Remember, a missorted input file is guaranteed to make comm fail like this. Ate you sure you had the same LOCALE during both sort and comm? If you use an exotic LOCALE, try setting your LOCALE to C.

And btw, uids are integers, perhaps those are user names.

QUOTE]And btw, uids are integers, perhaps those are user names.[/QUOTE]
uid stands for Unique ID, and as long as it is unique, it can consist of anything! A user name can also be a uid..[