[OTR-dev] Active fingerprint context pointer possible issue

[Ian Goldberg]

On Thu, Nov 22, 2012 at 09:29:26AM -0500, David Goulet wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA512
> 
> Hi Ian,
> 
> Yes I figured after sending the last email that it might be confusing
> on what I'm trying to do :). Here goes.
> 
> I'm currently working on the Irssi OTR module and I added a command to
> forget a given fingerprint. So either the user does the command within
> the chat window where for this case I have access to the OTR context
> so no need to pass a fingerprint to the command or the user specifies
> the human readable form of the fingerprint. (available fingerprints
> can be listed using an other command.)
> 
> With the later, the only thing I have is the string fingerprint (fp)
> so, for now, the only solution I came up with is to iterate over all
> OTR contexts of the user state for which I iterate over all fps and
> try to match the fp using the otrl_privkey_hash_to_human() and, on
> match, using otrl_context_find_fingerprint() for the fp object pointer.
> 
> Once getting this pointer, I have to check to see if this fp is NOT in
> an ENCRYPTED context before forgetting it.
> 
> Here lies my problem, the fp->context always point to the m_context
> which does not indicates me the right message state. For now, I've
> fixed my problem by returning back the context of the fp previously
> found with otrl_context_find_fingerprint() but I guess I should be
> able to access all associated context(s) of the fp and make sure it's
> safe to forget it? I guess, iterating on them starting with
> m_context->next ?
> 
> TBH, I'm just confused on how to proceed safely.
> 
> Hope this is clearer :)

OK, I think I understand.  What do you *want* to do if multiple children
of a single master context are using the same fp, and some are in
ENCRYPTED and some are not?

   - Ian