Apache Document Root Architecture

I agree with ahoffmann in theory, though I usually don't follow that advice in practice.  My current development environment uses the same Apache service, though I have several IPs to spread the love, so to speak.

You do need to make sure you have separate document roots, at the very least.  If your development application can be infiltrated, which is often the case before you have thoroughly vetted your code, it falls to Apache to keep isolation intact.

> .. it falls to Apache to keep isolation intact.
and the OS, and the DB, and other applications, and ... and ... and ...

theoretically you can secure apache and your application, and do you do that in practice, always, all day?

in theory, theory and praxis are identical, in praxis they're not

>>> and the OS, and the DB, and other applications, and ... and ... and ...

Too true.  If an application allows file access into the document root, then separating those document roots implements a needed level of isolation.

Of course, if the application allows unrestricted file system access (e.g., user-provided targets for file(), file_get_contents(), etc., without whitelisting), having separate document roots does not really do anything.  In that case, every document root would be accessible, limited only by file/directory permissions in the context of Apache's user.  But, that is something to be addressed in the application logic.

And I guess there lies the advantage of multiple Apache installations - you can set your production environment to use a much more restricted user than your dev environment.