Hacker News Reader

Comment by geertj | Hacker News Reader

Author here. Happy to answer any questions :)

I've never understood why password managers use browser plugins. It should be an application agnostic overlay that pastes selected passwords into the currently selected textbox.

I'm working around that at the moment with a combination of Emacs, the Mac keychain and Alfred.app - it's not beautiful though.

What are your plans for this?

FWIW, I think the browser plugins that 1Password implements are the "killer feature" - it notices you registering/logging-in to most websites, and prompts you to save the password. This reduces enormously the temptation to "just use an easy-to-remember reused-everywhere" password - I can randomly mash keys and have it automatically remember 25+ char passwords for me.

That _rocks_.

+1 to this. I would happily move to an open source password manager, but only if it implements this feature as well as 1p does

My plans regarding browser plugins was to have a small icon inside a <input type="password"> that can generate a new password, or would allow you to select the password if you already stored it and it's not clear which one to use. Otherwise the field should be automatically filled in.

How do you activate the paste in your setup? Is there some kind of global key combo for this?

As someone who is a big supporter of Firefox Sync [0]:

What will this provide me over Firefox Sync, which has a built in encryption layer on their (admittedly cloud) service? I can pull their code and host my own Firefox sync server, and someone has written a special PHP-based version for inclusion into OwnCloud. So, what is so special here except the use of buzzword P2P? (See next question below.)

How will you do to P2P without some kind of centralized bootstrapping service with the prevalence of NAT and people who do not know how to port-forward? (I just saw the Github repos, so I will examine this myself in a bit.)

UPDATE: It would appear you are using mDNS and/or Zeroconf [1] to handle this? I am still not sure how this will ease the NAT and/or syncing over public Internet.

Why should I trust you? What have you done other than this project? Not to be increasingly blunt, but you are not the Mozilla Foundation and your Github repos do not show much other than work on co-routines and AD service wrappers. That is useful to me, but I would like to see more security work and knowledge of crypto libraries if I trust you with password manager code. What else do you work on? Your blog does not show me too much, and I cannot find much mention of you to warrant 60,000 USD for such a project. I am correct in seeing you a Marketing/Product Manager? Sorry to be harsh, but honestly I think most of HN would be worried about these points.

UPDATE: Ok, maybe I was a little too harsh; you do seem to have some more experience than I had previously estimated, given some Googling and your Linkedin profile, if legit. [2]

[0] http://docs.services.mozilla.com/sync/ [1] https://github.com/geertj/bluepass/blob/master/bluepass/loca... [2] it.linkedin.com/in/geertj

I'm a big fan of Mozilla as well, but can Firefox Sync handle credentials other than for web? I use KeePass2 to store a lot more than just web username/passwords. Also, I like to keep my password database separate from my browser for security reasons.

As far as trusting this project, it is GPL and you are free to inspect it. I don't think being part of the Mozilla Foundation somehow guarantees secure code. If the author wants $60,000, he is free to ask for it.

Heh, well thanks for the honesty :)

> How will you do to P2P without some kind of centralized bootstrapping service with the prevalence of NAT and people who do not know how to port-forward? (I just saw the Github repos, so I will examine this myself in a bit.)

The P2P is done over local networks only. It uses multicast DNS for service discovery. Bluepass syncs between your own devices only. As long as the devices occasionally share a (W)LAN, it will work.

> Why should I trust you? What have you done other than this project? Not to be increasingly blunt, but you are not the Mozilla Foundation and your Github repos do not show much other than work on co-routines and AD service wrappers. That is useful to me, but I would like to see more security work and knowledge of crypto libraries if I trust you with password manager code.

Well, everybody has to start somewhere. I do believe that I have enough experience with crypto to be qualified here. I've got a physics degree and did a lot of maths. And I read Schneier and Wenbo Mao's book cover to cover. Also note that I'm not using any proprietary algorithms. This is all OpenSSL wrapped algo's that I'm using from Python.

But in the end, what should give this the right credibility is that the source code is up on Github, together with my intention to grow a community that can review the crypto. If there is something not right, which can always happen of course, then it will get out sooner or later.

> What else do you work on? Your blog does not show me too much, and I cannot find much mention of you to warrant 60,000 USD for such a project. I am correct in seeing you a Marketing/Product Manager? Sorry to be harsh, but honestly I think most of HN would be worried about these points.

This is not related to my day job. I would say have a look at the code, and decide for yourself if you like the code. I don't think being a product marketing manager in the day is mutually exclusive with being able to write good code.

The calculation of the funds is rather straightforward: $10k for each of the platforms that need work: iOS, Android, Chrome, FF, Mac and Windows.

>As long as the devices occasionally share a (W)LAN, it will work.

How about a VPN? I have a work laptop that is rarely (if ever) connected to my home network directly. -- However it is connected to my home network through a VPN daily.

I am not the network guy at my office, but if you look at his code it uses special DNS records (look-up Zeroconf, it is a discovery protocol for services on a network). Do you remember using Apple iTunes back when you could share music with people on the same network? Same difference, so to speak.

If your VPN provider, you or otherwise, does not block the mDNS/Zeroconf network traffic and had a permissible DNS configuration (almost all commodity routers have mDNS/Zeroconf config available if not on by default), I am sure you will have this work. The author can answer you more clearly.

It depends on how the VPN is setup... Multicast DNS will work if you are in the same L2 domain. Otherwise, a different solution is needed. For example, if you have a smartphone that you bring with you to the office, and that is connected to your office Wifi network and your home Wifi network, then synchronization would still work through your smartphone. Otherwise, a cloud-based "reflector" is needed. See the other comment about that.

And to apologize: re the reflector, I assumed you were doing this, like BitSync, and that is why I question ed you in my first comment. I am glad you do not over-promise with tyour project. Good work.

Thanks for the honest responses. I will not lie: you easily shredded most of my reservations. Not sure if I will donate just yet, but I will definitely consider it in the future.

And I did look at the code. It looks Pythonic and clean, so I am impressed. Definitely above my capabilities.

Regarding the question around Firefox sync: even if you use your own server, you are sending senstive data that is encrypted with a password or passphrase over the Internet. This opens you up to dictionary attacks now, and if the communications are intercepted, far into the future.

I confirm that [2] is my legitimate Linkedin profile :)

I have a KeePassX db in a git repo I clone on all my hosts plus two backup remotes.

Why would Bluepass be any better than my current setup? I already have P2P secure synchronization (via git+ssh), full control over my data and everything is based on portable FOSS.

A few things:

* Bluepass does push synchronization. So you wouldn't need to manually sync.

* It would work on smartphones / tables too.

* Are the remote git repos under your physical control? If not then your setup is vulnerable to a dictionary attack on your password / passphrase.

* The Bluepass database is set up such that it can resolve conflicts due to concurrent updates (actually it's an append-only graph of parent->child nodes with a algorithm that selects the most likely lineage in case of conflict).

[Edit: some clarifications]

I use keepass database that I store in a truecrypt volume which is then stored in dropbox which solves the first two items listed. With two separate 20+ char pass phrases the last issue is not one that I currently worry about.

This is not saying that this project isn't interesting, I've just found a solution that has solved these issues for me for the last several years.

I've been using a similar approach actually, with a keepass db on Dropbox. It provides a reasonable level of security assuming your passphrases are indeed random and computer generated (humans are lousy random number generators).

However I wanted to bring the security to the next level after that, and this is what Bluepass is about.

What advantages over 1password does your application give ?

For one thing - auditable source code. Except for my most paranoid moments, I pretty much trust AgileBits to not be selling me out to the NSA, but that'd be much easier to completely trust if the source code was public and audited/approved by smarter people than me who know their crypto shit.

Having said that, I'm currently storing my 1Password file in the unencrypted section of my DropBox storage, so my iOS devices can easily access it. (I've got EncFS/BoxCryptor working, but I don't think it's easy/possible to convince the DropBox app to read from the encrypted filesystem…)

I'm very excited to see this. I've been using KeePass2 on Linux and it's a pain to sync with my Android phone. I tried setting up OwnCloud to share the key file but even using that causes too much friction for me to be willing to frequently sync.

In your FAQ you say:

> "What do I get when I fund you? Once the software is ready, you will get unlock codes for the mobile versions for the amount you funded. As a special appreciation for being an early customer, these codes not expire and will unlock the mobile versions for life."

But you don't specify a price for the mobile versions. Also, will the mobile versions be GPL'ed?

Very happy to see you take Bitcoin as a funding option.

Hi, it's mentioned in the FAQ, little bit further down.

The mobile versions will be somewhere between $5 and $10. As a perk for funding Bluepass, whatever the price turns out to be, your version will include unlimited free upgrades.

The source of the mobile versions will be available, but for the success of a project, a recurring revenue stream is required to maintain them. The mobile platforms change a lot, and there are not many examples of open source projects that can successfully provide productized versions for these mass market consumer platforms. That is why I cannot commit at this point that the license will allow redistribution on an app store. This means it would not be an OSI style license.

Well, if you licensed the sources under the GPL, you could still prevent redistribution on iOS (though not Android), since the GPL is incompatible with Apple's ToS.

As the copyright holder, that wouldn't affect you, of course, since you don't need a license to distribute the software yourself.

I use keepass on PC and Android, and sychronise using bittorrent sync. No issues.

Just looked at bittorrent sync and all they offer is a binary download for Linux. Thanks, but I'd rather use open source tools to handle my secure data.

But given that my keepass file is encrypted and has a secure password - as I said, no issues.

Not a question, but a pet peeve. You have a typo on your homepage. Search for 'synchronziation'.

Also, please don't have text that looks like a hyperlink if it's not (even if it is a placeholder), specially if you then have hyperlinks that look the same.

Other than that, good luck on the project. It seems promising.

Any plans to support Yubikey for authentication? That alone would earn my backing.

Yubikey could be interesting for protecting the local vault. I haven't looked into it yet on how easy it is to do, but it is interesting and I will look into it. I cannot make any guarantees yet though.

Good project. Why aren't you using Kickstarter or Indiegogo?

Kickstarter is US/UK/Canada only. Being Dutch and resident in Italy it would not work.

I hadn't really looked at Indiegogo before I was too far along doing my own site. In the end it was relatively simple. I'll open source it at some point.

Have you considered supporting cloud sync? It should be possible to make it secure and it would be a huge convenience.

I did consider it, and the Bluepass sync protocol has an option for a node type without an encryption key. Such a node is merely "reflector" of traffic, which can be useful in case you have two disjoint sets of devices that are never on the same LAN.

This could be later addition. However it does decrease the security slightly because now you have to assume that others now have access to your encrypted data. It is still a lot better than the alternatives because still no dictionary attack is possible, you'd have to break RSA to get to it.

Does it support the keepass db format?

At this point it does not. But a feature to import a keepass DB into the Bluepass native format is in the plan.

So, does this mean that if i switch from keepass to Bluepass I won't be able to go back and will be stuck with a semi-commercial app forever?

Why not use keepass format natively? IMO a keepass-compatible app with built in sync would make much more sense. The userbase is already there, and a user friendly keepass app with sync would be a hit.

The Keepass format is not suitable for my use. Bluepass uses an append-only database containing a forest of nodes. Each node is a specific password version, and child nodes are updates of a parent. This data structure is used to do conflict resolution in case of concurrent updates. It also gives you infinite history, which is nice. The file format itself is an SQLite database with JSON documents in it.

Reply on news.ycombinator.com