New release on website

I have put the 0.3.0beta3 release on the http://multibit.org website:

20 Feb 2012 V0.3.0BETA3

Enhancements
+ Fix for network connection problems in 0.3.0beta2
+ Now tracks bitcoinj 0.4-snapshot instead of 0.3
+ I18n refresh, including Russian

Bugs
+ Private key import is currently not working correctly


New google groups mailing list: bitcoin-multibit

To make it a bit easier for people to keep up-to-date and discuss MultiBit, I have set up a Google groups mailing list:

bitcoin-multibit
http://groups.google.com/group/bitcoin-multibit?hl=en

Please join if you are interested in being part of MultiBit's progress.

thanks for the prompt and detailed response.  yes, i do run Windows so this will come in handy.  i'm obliged to check it out.  good work.

ok, i'm very impressed with this USB installation:

1.  do the private keys touch my RAM when sending/receiving?  doesn't the OS do the crypto signing mandating this?  security risk?
2.  is .0001 fee mandatory?
3.  i sent 13.83222962 Btc's to Multibit but only 13.8322 are displayed.  what happened to the 0.00002962 ?
4.  as i understand it, there is no RPC connection.  this is good for security.
5.  when will we be able to encrypt the wallets?

Hi cypherdoc,

Glad you got the USB install working ok.

Answers to your points inline:

1.  do the private keys touch my RAM when sending/receiving?  doesn't the OS do the crypto signing mandating this?  security risk?
Yes - at the moment the private keys are unencrypted in BOTH memory and in the wallets.   This obviously needs fixing before MultiBit can be used for larger amounts of BTC.   Encryption is pretty much at the top of my TODO list - needs to go into the underlying bitcoinj classes too.
   
2.  is .0001 fee mandatory?
It is a tricky one this as if I drop the mandatory minimum fee some people's BTC will get 'stuck' in the bitcoind nodes which would be painful all round.   I have gone with the same fee level/ solution as the bitcoind client for similar reasons.

3.  i sent 13.83222962 Btc's to Multibit but only 13.8322 are displayed.  what happened to the 0.00002962 ?
It is in your wallet - at the moment I have a fixed BTC number *display* down to 0.0001 BTC but the BTC in the wallet are in Satoshi. This is mainly not to overwhelm mainstream users with gazillions of numbers on the screen.
If you have a look in blockchain.info block explorer you should see the full amount really is there. 
Display formatting will I expect become more sophisticated (show in mBTC etc) over time but I do not have it as a very high priority at the moment.

4.  as i understand it, there is no RPC connection.  this is good for security.
Yes agreed.   I do not expect MultiBit to have any sort of programmable interface as it just creates an attack surface.   For an end client program it does not seem necessary.

5.  when will we be able to encrypt the wallets?
I plan to work on encrypting the private key output files hopefully in the next week or two.   The wallets will be next but will take a while because:
a) I want a common encryption format with the rest of bitcoinj
b) There is also a migration of the underlying wallet format (to protobuffers) which I will probably do first (Andreas has started doing it for his Android Wallet).   I don't want to change two things that are so critical simultaneously.
c) Testing testing testing.

I honestly do not know how long it will take.

6) Key output file location - by default it puts the export key file "next" to the wallet so you were probably using a wallet on your C drive.   If you hover your mouse over the name of the wallet in the 'your wallets' view it gives you the file path.   You can have wallets stored directly on your USB drive (wherever you like) and the default output private key file should be "next" to the wallet.

Note that the private key *import* on this release is broken at the moment.   You can import them into the wallets but the replay is faulty.   I am working on this.   You can import the whole key file into blockchain.info which works quite well.

I have been working on an "All-In-One" USB drive solution tonight that I have tentatively called "MultiBitThumb".
With a bit of work I think running MultiBit on a USB drive will simplify to:
a) You download a zip file.
b) You unzip it onto your USB drive
c) You then have links for "runOnWindows", "runOnMac", "runOnLinux" to run MultiBit locally with local wallets and blockchain on any of those OSes.

I will probably have to timebox this work - have a play with it tonight and then get back on bugs/ next steps tomorrow.

buggy.  can't get confirms to release that 12.8320.  i'm locked up here.  any suggestions?

Hmmm,

if you cannot close that open dialog window after a few tries you could kill MultiBit and then in the multibit.properties remove the line:
selectedView=XX

Start up MultiBit again and it should get rid of it.


RE: the amounts and confirms.
You can resync your wallet to what is on the blockchain which should get you back into a good state.
There is an option "Tools | Reset blockchain and transactions".

Read the notes and then run it and then leave MultiBit to resync.   (It is sort of doing a replay). Don't do anything whilst it is resyncing as it is literally replaying the blocks.

Of course you should not have to do any of this.


Edit: I am offline soon - if all else fails: try getting your BTC out by exporting the private key file and importing it into blockchain.info.   There is an option in "Import Export | Import wallet".  Then send it to wherever you want.    Sorry for your mixed experience with MultiBit tonight - it looks like I need to do some plain old fashioned testing/ bug fixing before adding more functionality.

With the minimum fee of 0.0001 a total of 0.0001 is stuck yes.  You cannot get it out without spending it all on fees.

RE: location of wallets.
Theorectically you can have your wallets and the blockchain that "feeds" it transactions anywhere but I think it makes it less error prone if the MultiBit exe, wallets and blockchain are all on the same USB.

At the moments the wallets do not keep track of the last block they saw so it works best if any given wallet is always sent transactions from a single blockchain.  We are beefing up the wallet format in bitcoinj and I am some plans on the block replay ability so this will improve.   You can imagine with the subtleties of reorgs that you don't really want a wallet to receive blocks/ txs from multiple sources though.


I think you have given MultiBit a good bashing this evening so thanks for the feedback.   It indicates a few areas that either need improving or more explanatory notes provided.

ok, if the .0001 is stuck thats fine.

but the export key fx is still confusing...

so i exported 5 private keys from MultiBit into Armory.  does it make sense that the 0.0001 btc left in the wallet would not be found in any of them?

wouldn't it be nice to have a option in the install process to choose between "normal" and "portable" installation?

Greetz!