Re: New feature?

[Ian M. <ian...@ea...>]

Michael,
Hmm ... you make very good points that I hadn't thought of.
I never realized authors had to put so much thought and energy into their 
creations.

Rehards,
Ian

On Saturday 05 June 2004 3:03 am, Michael Buesch wrote:
> On Saturday 05 June 2004 11:24, Ian MacGregor wrote:
> > However, I am now thinking of a another solution.
> > How does PwManager encrypt the file? Does PwManager use some system
> > specific info (ie. hard drive serial #, a random salt, etc.) as a key
> > when encrypting? Or, does it rely soley on the master password?
>
> The password-file is explicitely made portable, so that you can store it
> on your NFS server and access it from your main workstation and
> your notebook, for example. This would not work, when we depend on
> hardware.
>
> > You could have PwManager use
> > the masterpassword+hard drive serial number+random number to create a
> > grand master password and use that to encrypt with instead of just the
> > master password. that way a file encrypted on my machine would not
> > decrypt on any other machine due to the hard drive serial number and
> > random number being different.
>
> It may be an optional feature, but will _never_ be a good default
> behaviour, IMHO. People will complain: "Fscking PwManager corrupted my
> password file! I can't read it on my new Computer. I've sold my old
> one and I didn't make backups". Oh, yea. listen. I can hear tons of emails
> on their way... ;)
>
> > This would also defeat a keylogger and mouse mapper -
>
> I can't see how you can say that. Really.
> Secure file-format is the one thing, but entering the password
> is another. If we have this Hardware dependent file-format, there
> can _still_ be some key-loggers installed. Did I miss something?
>
> > and may be
> > easier to code.
>
> Oh, don't expect that.
>
> > But, I also could not use that file on any of my other three
> > computers. But, that would make it more secure.
> >
> > I am just trying to find ways to make PwManager more secure. It's a
> > wonderful app and I just want to see it improve.
>
> I want to see PwManager more secure, too.
> But I think there is one point we can't make more secure. => the user.
> All what we try to do here is to protect the user from himself.
> Imagine: What if we actually have implemented the hardware dependent
> file-format, a mouse and keylogger secure entering mechanism? What
> about a virus grepping through the memory image of PwManager? (/proc/PID)
>
> I think this is the same issue I discussed some time ago with George
> Staikos about KWallet. I complained that every application is able
> to be man-in-the-middle in the security critical KWallet DCOP data stream
> (and I fact that _is_ what PwManager does do emulate KWallet *g*. So
> consider PwManager as some kind of exploit. ;) )
> So he replied, that an application that wants to do Bad Things (tm)
> needs to run as the same user, as PwManager runs. And that's exactly
> protecting the user against himself.
>
> > Ian

-- 
Registered Linux User # 350412
MacGregor Despite Them!