Back to the code... :)

[grarpamp]

Hi :) Well the recent topic is really more for comp.risks
or some open source advocacy list, not here.

Seeing as how there's all sorts of people chiming in from out of
nowhere since 6mos for that one, lol... time to get back to the code...

How are things with the code these days?
Last I heard there was no formal license?
Is the spec done?
Anyone running nodes?
What's next codewise?
Who's on the development and maintenance ship or wants to be?
Where's the todo list?
Any bugs lodged to fix?
What do we have to do to take it and get it out
there and fire up some initial nodes for people?
Anyting else?

[Jeff Tyson]

Hi everyone,

 

On Thu, Mar 8, 2012 at 7:03 PM, grarpamp <grar...@gmail.com> wrote:

Hi :) Well the recent topic is really more for comp.risks
or some open source advocacy list, not here.

Yes, perhaps that was a troll just to get the communication flowing again :).

I think I can speak to at least a few of these questions as I have done work on the protocol. Basically, what I have done is allowed the DHT to work so that nodes can actually publish information about themselves to the network as a whole. In the process of implementing this, I have run into a painful but fixable bug related to the way the threads are used. I'm hoping to get this fixed with 3 to 4 weekends of coding and debugging. 

So this means that I *do* run nodes from time to time to debug. I do this in AWS (as it requires at least 20 separate machines). If anyone is interested in doing this on their own, I am more than willing to help guide them. Additionally, I've made some changes that should allow debugging with 20 processes rather than 20 machines but this still needs to be unblocked by the threading bug. 

Warm regards,

Jeff Tyson

 

Seeing as how there's all sorts of people chiming in from out of
nowhere since 6mos for that one, lol... time to get back to the code...

How are things with the code these days?
Last I heard there was no formal license?
Is the spec done?
Anyone running nodes?
What's next codewise?
Who's on the development and maintenance ship or wants to be?
Where's the todo list?
Any bugs lodged to fix?
What do we have to do to take it and get it out
there and fire up some initial nodes for people?
Anyting else?

--
You received this message because you are subscribed to the Google Groups "Phantom Protocol" group.
To post to this group, send email to phantom-...@googlegroups.com.
To unsubscribe from this group, send email to phantom-protoc...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/phantom-protocol?hl=en.

[grarpamp]

> Hi everyone,

> Yes, perhaps that was a troll just to get the communication
> flowing again :).

Nah, just someone foreign, and to the whole open source thing.
We love everybody though, especially our evil .se overlords :)

> I think I can speak to at least a few of these questions as I have done work
> on the protocol. Basically, what I have done is allowed the DHT to work so
> that nodes can actually publish information about themselves to the network
> as a whole.

Hmm, so we're not to the point of discovery when given a
seed node/list yet? Just the fixed nodes?

> In the process of implementing this, I have run into a painful
> but fixable bug related to the way the threads are used. I'm hoping to get
> this fixed with 3 to 4 weekends of coding and debugging.

Ok, cool.

On a tangent... how do people think about moving to git...
so maybe some of the various dev trees could become
collaborative and visible model? Plus having hash integrity
and the now commonality of git. github? maybe google?

> So this means that I *do* run nodes from time to time to debug.
> I do this in AWS (as it requires at least 20 separate machines).

I now see some mail from last year about a localhost array
and want to try that. Even FreeBSD jails would give simple
separate IP's, confs, etc in rfc1918 without VM overhead :)

> If anyone is interested in doing this [AWS] on their own

AWS and similar service is not free like localhost/rfc1918, right?

> Additionally, I've made some changes that should allow debugging with 20
> processes rather than 20 machines but this still needs to be unblocked by
> the threading bug.

Thanks for the notes from worker :) I do see some talk of Phantom
on i2p and tor forums so interest seems to be there and are awaiting
code that people can actually run and link up into a working test net.

[Ernst Rohlicek]

> > In the process of implementing this, I have run into a painful
> > but fixable bug related to the way the threads are used. I'm hoping to get
> > this fixed with 3 to 4 weekends of coding and debugging.
>
> Ok, cool.

+1

> On a tangent... how do people think about moving to git...
> so maybe some of the various dev trees could become
> collaborative and visible model? Plus having hash integrity
> and the now commonality of git. github? maybe google?

Well, the code is on a Google Code repo. I am sure Magnus is open to
accepting patches.

Of course, git allows for very nice distributed hacking and great
support for this kind of development workflow. Would certainly make
participation more accessible.

> Thanks for the notes from worker :) I do see some talk of Phantom
> on i2p and tor forums so interest seems to be there and are awaiting
> code that people can actually run and link up into a working test net.

I noticed this as well, it is also discussed on message boards in
Freenet and Phantom definitely is the next step in anonymizing P2P
networks, IMO.

I am not able to do big (working ;-) changes in C code, but
otherwise... ready and waiting :-)

> >> Last I heard there was no formal license?

Well, if I understand this correctly... Phantom currently is using the
HESSLA ?

If yes, why was this chosen?

I would like to remark that this license it not GPL-compatible (!) and
otherwise a pretty damn exotic license. The reasons against the HESSLA
are finely summarized here:

http://www.gnu.org/licenses/hessla.en.html

It is basically a GPL with "moral and purpose-of-use restrictions"
added, which are not enforcible anyway. For me personally, the GPL-
incompatibility is the knock-out argument, though. Makes it difficult
for people to build an ecosystem around this.

Phantom is supposed to be the foundation of a new generation - or even
new kind - of P2P anonymity programs and we should not limit its
usefulness.

Change to Apache 2.0 / MIT / LGPL or GPL would be recommended.


Greetings,
Ernst

[grarpamp]

> Well, the code is on a Google Code repo.

I meant I think google has a git system too, if not github.

> I am sure Magnus is open to accepting patches.

> Of course, git allows for very nice distributed hacking and great
> support for this kind of development workflow. Would certainly make
> participation more accessible.

Oh yes, didn't mean to imply lack of acceptance, contrary, things
seem good :) More about what it seems everybody uses now,
and a more flexible dev model/exchange. And hash/change integrity...
defeat evil google host company infectors :)

>> Thanks for the notes from worker :) I do see some talk of Phantom
>> on i2p and tor forums so interest seems to be there and are awaiting
>> code that people can actually run and link up into a working test net.
>
> I noticed this as well, it is also discussed on message boards in
> Freenet and Phantom definitely is the next step in anonymizing P2P
> networks, IMO.

Think the v6 interface is help here too :) Nobody else has native
IP interface.

> I am not able to do big (working ;-) changes in C code, but
> otherwise... ready and waiting :-)

Wonder what the code download stats look like so far?

> Well, if I understand this correctly... Phantom currently is using the
> HESSLA ?

I see now, yes, sortof. But be clear, there is both a spec license and a code
license. I think both are HESSLA. Or was it that the spec has not yet
been licensed? Which would of course put question the entire code so far.
I see no license in the whitepaper:
original_phantom_paper_with_updates.pdf
sha1 f28577cfbe2758588567aaf7012111bd7bc568f3

And the toplevel LICENSE file needs work. These are three separate things
referred to:

"This Phantom Protocol implementation"
"The Phantom Protocol specification"
"the original implementation of the Phantom Protocol"

But by positioning in the file "Licensed under the..." seems to
refer to just the spec. And both implementations are the
'original'.

So copyright and license each properly, "spec" and "this implementation".

> If yes, why was this chosen?
>
> I would like to remark that this license it not GPL-compatible (!) and
> otherwise a pretty damn exotic license. The reasons against the HESSLA
> are finely summarized here:
>
>  http://www.gnu.org/licenses/hessla.en.html
>
> It is basically a GPL with "moral and purpose-of-use restrictions"
> added, which are not enforcible anyway. For me personally, the GPL-
> incompatibility is the knock-out argument, though. Makes it difficult
> for people to build an ecosystem around this.
>
> Phantom is supposed to be the foundation of a new generation - or even
> new kind - of P2P anonymity programs and we should not limit its
> usefulness.
>
> Change to Apache 2.0 / MIT / LGPL or GPL would be recommended.

[https://www.gnu.org/licenses/license-list.html
there is not specific mit license, and lgpl/gpl mentions need versioned.]
Apache 2.0 is has unneeded non community provisions.

Well, yes, three clause mod BSD (or FreeBSD two clause) license is nice.
It's compatible with pretty much everything, and absurdly free.
The GPL crowd doesn't like BSD lets people 'steal' the code and use
it closed. Well, the open project doesn't really need the closed code giveback,
they will keep on coding past them in time. Users of code like Phantom need
to trust the code, so binary version unlikely ever win share, thus giveback
concerns really not enter as an issue. Which leaves 2/3 BSD.

PD (CC0/unlicense) is a nice giveaway.

Things like HESSLA is interesting, but there's no foundation, no
rich idealists in Phantom, so who's going to enforce it, what
jurisdiction, where the already test cases, etc? And the license is
really long, simple is less buggy and more readable.

Last, law routes around and trumps licenses. So all HESSLA's
activist principles could be voided by new laws, or permissions
in various jurisdictions. Doubt it stop much of anything, why
experiment?

Simple, common licenses such as BSD (or leftys/PD) seem allow routing
around things to occur most easily, and as the users choose best.
It's worth an evaluation.

There were at least some past list/blog about this. Maybe just not
enough people here yet to really give it due consideration.

[Ben Giles]

On Sunday, 11 March 2012 at 10:20 PM, grarpamp wrote:

I meant I think google has a git system too, if not github.

Google Code has Mercurial IIRC, which is a DVCS as well. That's the functionality I'm after. I prefer git, but whatever. Let the code branch (and merge) fast and often.

Simple, common licenses such as BSD (or leftys/PD) seem allow routing

around things to occur most easily, and as the users choose best.

It's worth an evaluation.

There were at least some past list/blog about this. Maybe just not

enough people here yet to really give it due consideration.

+1 for BSD, MIT or Apache over the GPL.

Actually, how about http://en.wikipedia.org/wiki/WTFPL

[grarpamp]

>> I meant I think google has a git system too, if not github.
>
> Google Code has Mercurial IIRC, which is a DVCS as well. That's the
> functionality I'm after. I prefer git, but whatever. Let the code branch
> (and merge) fast and often.

Were there a vote, same here, I'm putting time into learning git and moving
my stuff to it from CVS. I know silly reason, but I in part git, among all DVCS,
because it becoming big market share. Like CVS was 'the' vcs of late. So it
felt like good time invesment. Guess all the DVCS are similar tech, just
different front ends. So whatever.

> Simple, common licenses such as BSD (or leftys/PD) seem allow routing
> around things to occur most easily, and as the users choose best.
> It's worth an evaluation.
>
> There were at least some past list/blog about this. Maybe just not
> enough people here yet to really give it due consideration.
>
> +1 for BSD, MIT or Apache over the GPL.
>
> Actually, how about http://en.wikipedia.org/wiki/WTFPL

As with other minimal ones, it has nice do whatever spirit, whether
in 'wtfe' words or not. It's basically experimental PD, like beerware, etc.