Cookies

[David Bolen]

"Ken Seehof" <kens at sightreader.com> writes:

> Does anyone know of another mailing list more specific to http questions?
> 
> 1. Client downloads an execuatable (frozen python application).
> 2. Client downloads a web page.
> 3. Web page javascript generates a statistically unique random identifier.
> 4. Web page sends identifier to the server, which uses it as database key.
> 5. User runs python application
> 6. *** Python application somehow gets the identifier from somewhere ***
> 7. Now the python application and the server both have the same identifier.

Because of the restrictive client Javascript environment you'll
probably be faced with, crossing the boundary between inside the
browser and outside is not trivial.  

Is there a reason that the identifier has to only be generated by 3?
It's not clear if your Python application is going to connect to the
server anyway, but if so couldn't it just do the same processing
you're putting into the javascript and generate a unique identifier
(that it thus knows about) at that point?

Or, if it isn't going to connect to the server, how about just letting
the Python application open up it's own web page internally to run
appropriate code on your web server to generate the identifier (which
could run as a server script), and then Python could parse it right
from the resulting HTML information.

> The only question I am asking is how to do step 6.
> 
> Alternative:
> 
> 1. Client downloads an execuatable (frozen python application).
> 2. Client downloads a web page.
> 3. *** Web page javascript makes a unique identifier for the client. ***
> 4. Web page sends identifier to the server, which uses it as database key.
> 5. User runs python application
> 6. *** Python application somehow generates the same identifier ***
> 7. Now the python application and the server both have the same identifier.
> 
> Again the only question I am asking is how to do steps 3 and 6 to get the
> same number.

This would probably prove even harder.  Algorithmic generation of a
truly unique identifier will almost always include some time based
deviation, so I'm not sure about an algorithm that will both be
globally unique and re-computable at different times - at least not
without still passing some information between the two compute engines
which puts you back at your same problem.

--
-- David
-- 
/-----------------------------------------------------------------------\
 \               David Bolen            \   E-mail: db3l at fitlinxx.com  /
  |             FitLinxx, Inc.            \  Phone: (203) 708-5192    |
 /  860 Canal Street, Stamford, CT  06902   \  Fax: (203) 316-5150     \
\-----------------------------------------------------------------------/