On 24 Aug 2006 10:46:58 -0400, Rowland <[EMAIL PROTECTED]> wrote:
A me-too and a summary of the discussion thus far as I see it:

1. Breaking backward compatibility is a bad thing.
2. Saying you won't ever do it again is small comfort.
3. Providing a migration path would help a lot.
4. I don't care about the darknet. I don't object to its existence but I have 
no interest in it.
5. I want the opennet!
6. Backward compatibility between 0.5 and 0.7 looks like a foregone conclusion 
at this point.
7. What we need instead is a migration path from Freenet 0.5 to the 0.7 opennet.
8. And we need it badly. This could be a show stopper.
9. This should be a high priority item.

Now... what's the migration path gonna be?

First of all, it should be noted that there are significant technical
reasons that migrating content is difficult.  Key size changed from a
variable size from (IIRC) 1kB to 1MB to the current version of 32kB
for CHKs and 1kB for other keys.  Since most content needs bigger keys
than that, splitfiles are transparently supported.  The result is that
the locations of all the files changed.  Any data migration plan would
have to deal with this, and doing so is a non-trivial project.
Specifically, it would require more work on the part of the developers
(who are seriously overworked as it is) than it would require of the
users to reinsert the content on the new network and modify things as
needed.

1 -- yup, I agree.  But you were warned long ago it might happen.  It
did.  There were good reasons, and it wasn't done lightly.

2 -- actually, I don't think anyone has said it won't happen again.
Just that they will work very very hard to have it not happen before
1.0 gets released, which is still at least a couple years away.

3 -- yes, it would.  Please feel free to contribute one.  Personally,
as a donor to the freenet project, I would prefer my money go toward
doing cutting edge research into anonymity, and making the 0.7 version
work.

4 -- that's either naive or short-sighted.  No matter where you are,
it looks likely that opennet won't be a viable option forever.  We
can't just wait until that happens to produce an answer.  Also,
darknet has *inherent* security advantages that have been discussed
numerous times.

5 -- Yep.  So do most of us.  Myself included.  Getting enough refs is
pain when you don't know many people running Freenet (I know exactly
1).  Opennet is coming; be patient.  There are a lot of reasons not to
deploy opennet yet.  Most of them boil down to a) darknet is easer to
make work well and b) it doesn't yet.  Combined with c) most things
that improve the darknet performance will help the opennet, it seems
reasonable to try to fix the darknet first.

6 -- I think (and hope) it is.

7 -- We have one.  Install 0.7, and insert your content into it.  The
obvious improvement over this would be an easy way for most
applications to cross-post content.  In fact, I'll bet all most people
really need is for Frost to do that, and they can handle freesites etc
manually.  So please talk to the Frost devs about that, it would be
*way* easier than trying to solve the problem at the node level.

8 -- Doesn't appear to be yet.  It looks like the darknet is alive and
growing, though small.

9 -- I disagree.  If you want a file sharing program, go find one.  If
you want a chat program, go find one.  If you want an answer for
communicating in an actively hostile environment with legal backing to
its hostility, you need Freenet.  Except that you don't need 0.5,
because that doesn't solve the problem.  And you don't even really
need 0.7, because it doesn't yet either.  What you need is for smart,
motivated people to be working hard at finding a solution for you,
because AFAICT it doesn't exist yet.  And I think that's exactly what
the devs are doing.  Working on back compatibility would be a huge
drain on those resources, and they're working against a very real
clock.

Evan
_______________________________________________
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:[EMAIL PROTECTED]

Reply via email to