Hi,

Martin Duerst wrote in [ruby-core:18619]:
At Tue, 16 Sep 2008 19:37:30 +0900,
> Is the reason this has to be done with "make up" just that
> it's otherwise impossible to get a reliable (e.g. everything
> is really up to this revision) revision number?

Actually, if revision.h didn't exist it will be created.

It's not possible for all make to `run the command every time, but it
may or may not update the target'.

> Conceptually, it feels to me that updating should be
> separate/separable from the make process, and the make process
> should make sure that it has the right revision number.

Does `separate' mean a command independent from make?

The revision number of the working copy changes only when you did `svn
up', so the single target which does it with updating revision.h feels
reasonable.

-- 
Nobu Nakada