Access control - author authenticity using a shared integration repository

[Stanimir Stamenkov]

I'm evaluating various aspects of using Mercurial in the workflow we
employ in my company.  So far I'm trilled about the ease Mercurial
integrates into our workflow and that it actually corrects some
shortcomings we currently have.

This is something I haven't figured out, yet.  In our flow we have
an integration repository (currently a Subversion branch) where all
developers check-in their changes to tasks they have completed.  We
have very few release managers so it is not feasible for them to
pull and review every change on their own - code reviews happen
earlier in the process and they are not necessarily linked to VCS
check-ins.  I want to ask if anyone employs such an integration
repository where all developers have push access and how he/she/you
control the following case.

Often a task is worked on by couple of developers cooperatively.
The final product would contain changesets of all the authors which
have done the changes, but probably (and I don't see another option)
the push to the integration repository will be done by a single
person.  So it should be possible for a user to push changesets of
other authors.  In this scenario I don't see anything preventing a
user to forge a changeset with the credentials of another.  How do
you deal with this?  Is there trace of the push operations - who
have done and what changesets have been added with them?

-- 
Stanimir