Hacker News new | past | comments | ask | show | jobs | submit login
GNUnet 0.10.0 released (gnunet.org)
109 points by bratao on Dec 24, 2013 | hide | past | favorite | 28 comments



I always think it's a shame that they don't release binaries. Yeah sure, it technically "works on GNU/Linux, FreeBSD, OS X and W32", but without binaries, they'll barely get more than a few users. I really like that they continue to improve the technological underpinnings of the software, but there's a reason a lot of people use the tor network. There is a bundle that they can just download, start and it helps them connect.

I guess they DO see themselves as a "framework", but there is a GTK UI and there is a "regular" filesharing component to it


Releasing binaries is a lot more work. You can tell other people to do the work, sure, but this telling is unlikely to accomplish more than some doing.

Consider making a donation towards binaries, perhaps of build servers or Windows licenses so that this can be done. GNU projects are funded by the FSF, so a donation to the FSF along with a request for binaries carries a lot of weight.

Of course, the most weight is carried by a code contribution. Help GNUnet make those binaries!


Mac and W32 installers are coming soon.


Also notice how bitcoin and litecoin were introduced with a great gui. They sure picked up.


I wonder whether they would consider replacing GNS (GNU Name System, DNS substitute, decentralized using a chain of trust model but without globally unique names) with Namecoin (uses a blockchain to enforce a registration fee). The latter isn't perfect, but it's probably better, and I wish it had more uptake in general.


Sorry, but Namecoin is a scamcoin! DNS works on low-end computers, Namecoin is an overkill requiring vast storage and constant syncing - you can recall what happened a month ago when Coinbase went out of sync with the blockchain and were pretty much offline of hours. Using the blockchain for absolutely everything is an abuse! DNSSEC + DNSCrypt are pretty fine for me! Plus, being a scamcoin makes .bit domains pretty expensive - right now it's $14/yeah - way more than a way more useful .com!


I think that's unfeasible. Namecoin requires everyone to have a copy of all domain names AFAIK. It will just grow to the point of being useless if it becomes popular. If it becomes too big, you have to trust a small number of nodes to run a full instance and give you correct results. That's about as good as the current PKI/DNS system anyway.


Even if you do have to use a server, it's a lot better to be able to choose to trust anyone with a terabyte hard disk (with about 150 million current domains according to whois.sc, that sounds like enough even if it gets very popular) to not return incorrect data for a fundamentally decentralized store than to trust each of a large group of entities (CAs) to uphold a system whose canonical source (DNS) is subject to censorship anyway.


There are Bitcoin wallet implementations that don't require the full block chain. I suspect a lightweight version of Namecoin is also possible.


I wonder when GNUnet and GNU Social will be eventually combined.


I don't think they will. I would like to see work in making GNU social work over GNUnet though.

If anything, GNU FM and GNU social will work better together.


"GNUnet used RSA 2048 since its inception in 2001, but as of GNUnet 0.10.0, we are "powered by Curve25519"

anything to do with recent RSA news?


What recent RSA news? The Reuters thing? RSA-the-algorithm has nothing whatsoever to do with RSA-the-company.

(It is good that they are moving away from RSA; new systems should stop using RSA.)


Why should new systems avoid RSA?


Because it's slower and has larger keys for equivalent security levels. There's growing concern based on recent advances that RSA and DH may not be as secure as is generally thought. [1]

Despite the unease certain people (like Schneier) have with ECC in general (it is more advanced math and fewer people understand it well), and NIST curves in particular (did NSA choose weak classes of curves?)-- based on public knowledge, there are advances in factoring and the discrete log problem, while there's no similar progress against ECDLP.

If NSA knew some attack against ECC or weakness in p256, p384, p521 curves (i.e. anything which doesn't apply to RSA), that would mean they sabotaged the security of Secret and TS information that's allowed to be transmitted using NIST suite B, which uses ECC and specifically those curves. Military comms between NATO allies might also be using Suite B.

Is it possible that NSA has gone completely off the rails, only cares about the security of their own ECI (which I think tends to use their Suite A algorithms)? Are they willing to throw all other cryptography users, including NATO military comms users and other parts of the US Government, under a bus in pursuit of their ability to weaken generally used encryption? Perhaps, but I think it's unlikely. Speaking of which, does GCHQ use Suite A algorithms, or Suite B (including standard ECC and NIST curves), or something else? And the rest of the Five Eyes? That's something that unreleased Snowden documents probably speak to.

[1] http://www.slideshare.net/astamos/bh-slides


Yep.


Marketing.


GNUsenet seems like a more appropriate name.


Zero dot tEn dott zero..?


Can you GNU guys maybe stop putting GNU in front of everything? It's stupid. It makes me want to avoid these products. The message it sends is that the licence is more important than the product.


GNU License ≠ GNU Project. Yes, (almost) all GNU software (as in projects of the GNU Project) are released under the GPL. So is a lot of other software.

GNU Foo is, branding-wise, just like Microsoft Word, Google Drive, or Apple Macintosh.


Okay, so brand fail. I've using linux since my first lead dev introduced me to it back in '97-98, and it used to take a weekend to get set up right, and the only help on the newsgroups or BBS's was to RTFM. So maybe the brand could work for others, but I have not seen any really positive images of the brand. Ugly Gnu ink mascot, Restrictive GPL Licence that is less about sharing and more about control, Richard Stallman unable to give it a rest. The brand isn't a good one.


Since you are new, I will give you some benefit of doubt and provide a portion of the Hacker news guidelines:

"Please avoid introducing classic flamewar topics unless you have something genuinely new to say about them." - http://ycombinator.com/newsguidelines.html

If you feel that a license that says "Full permission, so long you do not restrict other users of the program" as more about control and less about sharing, then keep that to you self. There is nothing genuinely new about in such discussion, and frankly, it just tiresome to read, comment and down vote. Note that the down vote is not because I 100% completely disagree with your comment, but because your comment will just cause a rehash of an flame war which is old enough to start investing in its own pension plan.


It would only constitute a flamewar topic if he were advocating an alternative to GNU.


Quote: "Restrictive GPL Licence that is less about sharing and more about control"

If that aint an old flamewar statement, I don't know what is. Everybody here knows what the alternative is (BSD, and BSD-like licenses). We weren't born yesterday and we're nerdy enough to know what is being argued for.


The license is a tool. The importance is the freedom. We work on a branding, the GNU brand.


Sometimes the license is more important than the product.


The licence may be more important a product, but the licence attached to a product is not more important than the product.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: