> I don't think it is reasonable to put a myriad of command-line=20 > switches into an app to accommodate for every possible use of the=20 > software. "Feature creep" has been a subject of lots of very lengthy=20 > debates, and the consensus has been that the more "special cases"=20 > there are for a program, the harder it becomes to maintain it,=20 > particularly when a need arises to rewrite significant chunks of code. In general, I agree. As I said, it's easy enough to work around, if slow. > Besides, what you ask is counter-intuitive. Consider this yourself: >=20 > yum update pkg1 pkg2 pkg3 >=20 > This is an atomic request for one transaction -- update three=20 > packages. There can be two possible outcomes of this command: >=20 > a) Success: all three packages have been updated > b) Failure: neither of the packages have been updated Except I read this command line as "make sure I'm running the most recent version of all of these packages". So if it doesn't do that, I'd expect it to return failure. So I expect the outcome to be: c) Success: all three packages are the most recent version Your argument is like saying the SQL statement: update table set value1=3D'foo', value2=3D'bar' must fail if value1 is already 'foo'. Eric
