Hi Does someone know how Unix or Linux implement
'diff' command ? Any good reference or the
algorithm will be better ?

Thanks

Well, I assumed you read the man page...
man diff

...otherwise, read...
http://unixhelp.ed.ac.uk/CGI/man-cgi?diff

For example:

File1
1col_1,1col_2,1col_3,1col_4,1col5
2col_1,2col_2,2col_3,2col_4,2col5
3col_1,3col_2,3col_3,3col_4,3col5
4col_1,4col_2,4col_3,4col_4,4col5
5col_1,5col_2,5col_3,5col_4,5col5

File2
1col_1,1col_2,1col_3,1col_4,1col5
2col_1,2col_2,2col_3,2col_4,2col5
0col_1,3col_2,3col_3,3col_4,3col5
4col_1,4col_2,4col_3,4col_4,4col5
5col_1,0col_2,5col_3,5col_4,5col5

diff File1 File2

3c3
< 3col_1,3col_2,3col_3,3col_4,3col5
---
> 0col_1,3col_2,3col_3,3col_4,3col5
5c5
< 5col_1,5col_2,5col_3,5col_4,5col5
---
> 5col_1,0col_2,5col_3,5col_4,5col5

...lines starting with "<" belong to the first file (File1).
lines starting with ">" belong to the second file (File2).

Is there somthing specific you are trying to do with diff?

Thank you so much.

Because currently we have two programmers working on one file. I am just wondering how to combine their work daily. And I know in CVS they use
'diff' to compare different version's file. Or maybe you have any good suggestions?

Thanks

Normally, CVS can be used to "check out for edit" source modules
and prevent (in an advisory manner) more than 1 person
from making changes to a single module at the same time.
This way, changes made by one person can safely be made
are are sure to be in the "check out for edit" by the next person.
If it is necessary to have more than one person working on
the same source module (and in some cases it is - but rarely)
you can utilize the CVS "merge" function to merge the changes
from both files into one. May I suggest your developers
familiarize themselves with the CVS docs...

http://www.gnu.org/manual/cvs/index.html

...it will benifit them greatly.
The CVS "diff" and UNIX "diff" are two different things.
I was describing the UNIX command diff(1).