Freedom Box | Bitcointalksearch.org

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: 2112 on October 26, 2011, 04:02:29 PM

Quote from: DeathAndTaxes on October 26, 2011, 03:17:58 PM

The entire software wouldn't have to be written just the database. It wouldn't be a trivial task but it could be done. Alternatively you could use a 1.7" hard drive but those are going the way of dinosours as flash gets cheaper and cheaper. Maybe possibly do some kind of caching to reduce number of writes required by the database. I am sure there is a solution.

I fully agree that it could be done. After a rewrite Bitcoin's blockchain is actually an ideal application for the flash storage: only appends, no re-writes, massive reads. So no wear-leveling and any filesystem would be required for the storage of the biggest item: the block-chain itself.

But the combination required: (1) rewrite of the Bitcoin client (2) dedicated raw flash storage (3) suitable modifications of Linux kernel to expose (2) to (1); make this whole project unlikely.

I don't see how 2 & 3 are an issue. I am talking about using off the self SD/SDHC cards or USB thumbdrive. Linux supports mounting partitions on flash currently.

2112

legendary

Activity: 2128

Merit: 1073

Quote from: DeathAndTaxes on October 26, 2011, 03:17:58 PM

The entire software wouldn't have to be written just the database. It wouldn't be a trivial task but it could be done. Alternatively you could use a 1.7" hard drive but those are going the way of dinosours as flash gets cheaper and cheaper. Maybe possibly do some kind of caching to reduce number of writes required by the database. I am sure there is a solution.

I fully agree that it could be done. After a rewrite Bitcoin's blockchain is actually an ideal application for the flash storage: only appends, no re-writes, massive reads. So no wear-leveling and any filesystem would be required for the storage of the biggest item: the block-chain itself.

But the combination required: (1) rewrite of the Bitcoin client (2) dedicated raw flash storage (3) suitable modifications of Linux kernel to expose (2) to (1); make this whole project unlikely.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: 2112 on October 26, 2011, 03:04:45 PM

Quote from: DeathAndTaxes on October 26, 2011, 02:56:03 PM

The OS, client, wallet, and blockchain can fit on a 4GB SD card. Eventually you will wear it out but who cares. Flash storage is incredibly cheap today.

Not with Berkeley DB used for index logging. The write-ahead-logging is the worst case for the wear-levelers: almost-all writes and almost-no reads. Complete software redesign would be required. And to make a redesigned software bug-compatible with the official Satoshi client is an exercise in pointlessness.

The entire software wouldn't have to be written just the database. It wouldn't be a trivial task but it could be done. Alternatively you could use a 1.7" hard drive but those are going the way of dinosours as flash gets cheaper and cheaper. Maybe possibly do some kind of caching to reduce number of writes required by the database. I am sure there is a solution.

2112

legendary

Activity: 2128

Merit: 1073

Quote from: DeathAndTaxes on October 26, 2011, 02:56:03 PM

The OS, client, wallet, and blockchain can fit on a 4GB SD card. Eventually you will wear it out but who cares. Flash storage is incredibly cheap today.

Not with Berkeley DB used for index logging. The write-ahead-logging is the worst case for the wear-levelers: almost-all writes and almost-no reads. Complete software redesign would be required. And to make a redesigned software bug-compatible with the official Satoshi client is an exercise in pointlessness.

2112

legendary

Activity: 2128

Merit: 1073

Quote from: Red on October 26, 2011, 02:50:42 PM

You could easily do a light version that did not try to validate every transaction.

That is what I'm saying. You could do all kind of modifications to the original Bitcoin P2P protocol and see if it can survive. Original design was really Peer To Peer. With modification it will become Parasite to Parasite.

The question still stands: what is the ratio of Parasites to Peers that the overall bitcoin protocol can still tolerate?

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: 2112 on October 26, 2011, 02:41:11 PM

Quote from: doobadoo on October 26, 2011, 02:05:25 PM

dis
The idea is to run a debian OS on a cheap ARM SoC that plugs into your wall.

Doesn't have a disk big enough. Without the copy of the blockchain all the lite-clients could be just blindly circulating the invalid transactions. For now the vast majority of clients are able to consult the blockchain and suppress the invalid transactions.

In a way, this could be a good test of the resilience of the bitcoin P2P protocol.

The OS, client, wallet, and blockchain can fit on a 4GB SD card. Eventually you will wear it out but who cares. Flash storage is incredibly cheap today.

Red

full member

Activity: 210

Merit: 115

Quote from: 2112 on October 26, 2011, 02:41:11 PM

Doesn't have a disk big enough. Without the copy of the blockchain all the lite-clients could be just blindly circulating the invalid transactions. For now the vast majority of clients are able to consult the blockchain and suppress the invalid transactions.

You could easily do a light version that did not try to validate every transaction. After all, a client like this couldn't really generate blocks. It would burn itself up and shut down all other services trying to calculate proof-of-works.

All at freedom box bitcoin app would have to do is to receive the latest block in the chain and look for wallet related transactions. It could then generate and send new transactions from a simplified web client. In every case though, it should store bitcoin private keys off of the plug.

2112

legendary

Activity: 2128

Merit: 1073

Quote from: doobadoo on October 26, 2011, 02:05:25 PM

dis
The idea is to run a debian OS on a cheap ARM SoC that plugs into your wall.

Doesn't have a disk big enough. Without the copy of the blockchain all the lite-clients could be just blindly circulating the invalid transactions. For now the vast majority of clients are able to consult the blockchain and suppress the invalid transactions.

In a way, this could be a good test of the resilience of the bitcoin P2P protocol.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

As soon as someone writes it. The client code is open source. Anyone can modify it for alternate purposes/uses.

Interesting idea though. A secure web front end for hardware client.

doobadoo

sr. member

Activity: 364

Merit: 250

great so when does bitcoin get an SSL HTML network interface like many bittorrent apps? Just load a wallet up on your freedom box, then use an SSL browser to access your client, say from your phone. Maybe THAT'S the way for a lite client, ala android/ios devices.

As soon as MyBitcoin.com releases their source i guess. ;-o

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Debian is a "flavor" of Linux. No need to port anything. There already is a bitcoin linux client.

doobadoo

sr. member

Activity: 364

Merit: 250

Any one familiar with Eben Moglen and his whole "Freedom Box" initiative? The idea is to run a debian OS on a cheap ARM SoC that plugs into your wall. Shouldn't this device have a bitcoin client on it. The idea is to make a cheap and easy to use server, and implement software that will allow from encrypted/anonymized P2P communication.

Maybe it can also contain a bitcoin app in the bundle? It would need to be ported to Debian, which is the OS they're using. I guess one would interface via an SSL web session? Thoughts every one?

(It appears that they also accept donations in bitcoin: 1MyZjxnLgun6APrDkkh7ffrQJyy6xbuDho)
https://www.freedomboxfoundation.org/donate/

Topic: Freedom Box (Read 1218 times)