[Rubygems-developers] More work on the remote installer

Jim Weirich jim at weirichhouse.org
Wed Dec 1 09:49:09 EST 2004

On Wednesday 01 December 2004 09:25 am, Hugh Sasse Staff Elec Eng wrote:
> > (B) Add support for a writable per-user source info cache for use when
> > the global cache cannot be written.j
> Nobody has responded to my remarks about possible denial of service
> attacks by users if data from this is merged with the system cache.
> Have I misunderstood this?  Is there no way a user can damage the
> site-wide data if they choose?

My plan is to update the site-wide file if the current user has write access 
to it.  Otherwise update the user specific file.  Both of these updates would 
be from the downloaded information.  I would not update the site-wide file 
with info from the user.

However, the user could specifically update his own copy manually.  I'm not 
sure I understand your concern over a DOS (Denial of Service ... right?) 
attack based on a hacked cache file.  These cache files are merely hashes 
that map source and gem names into gemspecs for searching purposes.  If the 
cache file is incorrect, it would mean that the gem command would attempt to 
grab a version of a gem from the server that might not be there (or tell you 
it can't find a gem that really is there)  Could you expand on your concerns 
about DOS attacks.  Thanks.

> http://www.theregister.co.uk/2004/11/29/ie_security_holes/
> I'm trying to think of what is possible when you can only really
> trust some users on a system.  There's a reason why ruby and other
> languages need a root install....

-- Jim Weirich    jim at weirichhouse.org     http://onestepback.org
"Beware of bugs in the above code; I have only proved it correct, 
not tried it." -- Donald Knuth (in a memo to Peter van Emde Boas)

More information about the Rubygems-developers mailing list