This is a different issue than obfuscating data. You are talking  
about making a webapp that doesn't trust the server. The Web (as  
defined as HTML+JS+CSS+HTTP+etc) is not designed to provide this kind  
of end-to-end security, and attempts to layer it on top are fraught  
with peril. Solutions to this problem are numerous and varied, but  
they will require a better underlying system.

--Noah

On Jul 31, 2007, at 7:30 PM, Vincent wrote:

>
> Oh, and in this case, it's an internal wiki behind a VPN+firewall.
> Access and making sure that the information isn't stored in a
> compromised format is the issue as I understand it.
>
> On Jul 31, 3:52 pm, Noah Kantrowitz <[EMAIL PROTECTED]> wrote:
>> This feature really makes little sense. The correct way to do this is
>> server-side authorization controls. SSL provides transport security
>> already. The "once the users password is cracked" issue is
>> orthogonal, and means you need a better authentication system than
>> simple passwords. I would recommend a two-factor system, and I know
>> RSA's SecurID can be integrated with Apache.
>>
>> --Noah
>>
>> On Jul 31, 2007, at 6:40 PM, Vincent wrote:
>>
>>
>>
>>> I would tend to agree with you.  In fact I completely agreed with  
>>> you
>>> until I looked into this a bit.  And yes, ultimately, you're kind of
>>> right - end-to-end security and HTML isn't quite consistent.  But in
>>> the event of hey I want to post a message and provide an easy way to
>>> encrypt/decrypt it if I have the key actually seems reasonable to  
>>> me.
>>> It prevents data from being stored in the clear in the wiki or in a
>>> file.  In fact, it's out in the open - it's the encryption key that
>>> has the be known.
>>
>>> In this case, the context-dependent permissions system isn't what is
>>> desired.  The client doesn't want the permission to be user-specific
>>> as once the user password is cracked, then it's all over.  It's a  
>>> 2nd
>>> level of security that can be applied to a given wiki page.  At  
>>> first
>>> I thought .htaccess or something like that would be enough, but he
>>> emphasized that he didn't want the data to be stored unencrypted
>>> either, hence the reference to the Rails app below/above.
>>
>>> Make sense?  Still haven't checked into PGP meets Javascript, but it
>>> seems like it would be trivial to do a:
>>
>>> text=dom.text
>>> cryptkey = dom.cryptkey
>>> newtext = togglecrypt(text, cryptkey);
>>
>>> So... while I was typing I did a bit of research and found the
>>> following page:
>>
>>> http://www.fourmilab.ch/javascrypt/jscrypt.html
>>
>>> This pretty much does what is needed, so I guess the next step would
>>> be to see what it would take to modify the Wiki edit form to include
>>> an encrypt using this key button.  And the Wiki view page to include
>>> an decrypt using this key button (form) that called these functions,
>>> possibly adding some settings in trac.ini so that you could control
>>> which kind of encryption/decryption cipher you would be using (i.e.
>>> base64, hex, blah)
>>
>>> Make sense?  Is this something I should submit as a plugin or try to
>>> work it out as a feature that can be enabled/disabled in trunk?   
>>> Would
>>> be curious if anyone else on the mailing list would find this  
>>> useful,
>>> with the idea that you can now store your passwords on your wiki,
>>> presuming that your encryption key is secure w/o worrying about even
>>> those pesky boys who manage to get root access to your wiki and
>>> database... they still can't read the text without the key.
>>
>>> Vincent
>>
>>> On Jul 31, 3:01 pm, Noah Kantrowitz <[EMAIL PROTECTED]> wrote:
>>>> You should use the new context-dependent permissions system.  
>>>> Look at
>>>> the authz-policy sample plugin. If you want end-to-end security,
>>>> don't use HTML.
>>
>>>> --Noah
>>
>>>> On Jul 31, 2007, at 5:47 PM, Vincent wrote:
>>
>>>>> As an addendum, the client who requested this feature found this
>>>>> link
>>>>> to an equivalent feature written in Ruby.  Seems to me this  
>>>>> could be
>>>>> done client side (presuming there is some sort of cryptology  
>>>>> library
>>>>> available for javascript) but it could also be done as a python
>>>>> process.  Will do more research and get back to the group.
>>
>>>>> Vincent
>>
>>>>> On Jul 31, 2:01 pm, cobwebsmasher <[EMAIL PROTECTED]> wrote:
>>>>>> This is something I hadn't thought would be useful, as I think
>>>>>> there are plenty of workarounds to get a similar effect, but I
>>>>>> wanted to know what the group thought of this idea:
>>
>>>>>> Encryptable Wiki pages that can only be unlocked by either a
>>>>>> specific password and/or a specific username/password  
>>>>>> combination.
>>
>>>>>> Client wants a place to store usernames/passwords, but doesn't
>>>>>> want to store it on a wiki as you could work around the wiki to
>>>>>> get the information.  I've suggested just placing a zip file with
>>>>>> a password as an attachment on the wiki, but he really wants this
>>>>>> notion of an encryptable wiki --- it's stored in the database
>>>>>> encrypted so that only users with a specific password can read
>>>>>> what it's saying.
>>
>>>>>> I guess you could rig up an AJAX DOM sweep that took a given
>>>>>> document through some form of reversible encryption, but I  
>>>>>> have no
>>>>>> idea how secure one can make any reversible encryption scheme if,
>>>>>> at the end of the day, a password is involved.
>>
>>>>>> Thoughts on the idea and possible implementation?
>>
>>>>>> Thanks,
>>
>>>>>> Vincent
>>
>>>>>> ---------------------------------
>>>>>> Got a little couch potato?
>>>>>> Check out fun summer activities for kids.
>
>
> >
>


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Trac 
Users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/trac-users?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to