[Chicago] URL + authentication
kumar.mcmillan at gmail.com
Thu Jun 19 04:59:26 CEST 2008
This sounds similar to what Tahoe does and it is written in Python
(probably with some C). You could use Tahoe as a backend or steal
parts of their code. Tahoe uses Erasure coding for storing the data
but I'm not sure how it exactly sends data over the wire. It is
designed for these kinds of segmented sharing groups you are talking
about and is nice because it is distributed.
Although FWIW I can't get http://allmydata.com/ (a Tahoe based backup
service) to work very well, performance wise. Not sure what that's
about but I gave up after a while.
On Wed, Jun 18, 2008 at 9:20 PM, Michael Tobis <mtobis at gmail.com> wrote:
> So suppose there is some proprietary data. A lot of it. Lets say gigs of
> binary. Per file. Maybe thousands of multi-gig files per resource pool,
> maybe dozens of such pools. Nevertheless, the members of one pool don't
> necessarily want the members of another pool swimming in their own pool
> Nevertheless, there is a service (the code is open source in actual fact but
> that doesn't matter for these purposes; the point is that there is a
> service) which allows access to all of these pools subject to strong
> I want to create clean code that, if I am authenticated with respect to a
> specific resource pool, should allow me to have remote read access. I just
> want to know how to keep the security from messing with the elegance of the
> code. So.
> Does anyone have a good suggestion for a very clean representation of a
> secured remote object?
> I'm guessing the authentication will be a method of the object, and it will
> have a URL as an attribute. Some sockets and encryption would help.
> Is this a solved problem or do I need to put it together for myself?
> Oh, did I mention Python? Oh bother. Sorry. Python, python, python!
> Chicago mailing list
> Chicago at python.org
More information about the Chicago