John Plocher wrote:

> Other than those specific activities, I fail to see how much of anything 
> else one could do on the webapp would directly affect the source,
> since all source code modifications go thru ssh/svn/hg and NOT thru the
> webapp.

It's an aphorism that any security system is only as strong as the 
weakest link.

> Think about it.  I could make the OS.o web site 100% secure in about 30
> days.  At the end of the period, I could guarantee that the site was
> completely secure - that there was no way for malicious users to impact
> anything sensitive on the site.  All I would have to do is call up the
> OS.o hosting provider, cancel its hosting contract, cancel the DNS domain
> registration and unplug the machines (etc).  100% secure, and 100% 
> unusable.

Umm, I'm not sure what relevance that has, as that's now what is being 
proposed.

> This shows that security is not the only goal of a system; usability is 
> important as well.  If users can't do their tasks easily, they will
> leave, or the tasks simply won't get done.
> 
> In this situation, rather than using a sweeping generalization (all users
> must revalidate after 4 hours of inactivity), a more usable approach might
> be to say all users must revalidate after 2 weeks of activity or 
> inactivity,
> as well as whenever they perform actions that affect the source trees or
> other user's roles.  These latter revalidations might well have a 1 to 4 
> hour
> rememberme cache associated with them to make their use more palpable.

Two weeks is too long.  sun.com uses an 8 hour inactivity timeout, with 
a enforced re-login after 24 hours.  For OSO an 8 hour inactivity 
timeout with no enforced re-login seems reasonable.

-- 
Alan Burlison
--
_______________________________________________
website-discuss mailing list
[email protected]

Reply via email to