Re: Rename atomicity

[Paul Eggert]

> From: Assar Westerlund <address@hidden>
> Date: 06 Mar 2001 17:51:45 +0100

> If you're writing code that
> depends on this and that has to be able to run on systems where you
> cannot guarantee that rename isn't atomic, atomicity has to be
> accomplished in some other way.

Agreed, but I've been relying on atomic rename for over a decade in my
code and in practice, I've found this to be a pretty safe assumption.
For example, RCS has been using rename for atomicity since 1983, and
I've never gotten a bug report about it.

I occasionally hear of POSIX-like systems where rename isn't atomic,
but on those systems my impression is that nothing is reliably atomic,
so there's not much point trying to cater to them if your goal is
reliable atomicity.

The situation was different 20 years ago before rename became popular,
but those ancient systems are no longer of practical concern.