Armory - Discussion Thread

[etotheipi]

Initial Testing Release of Armory (0.72-beta)-alpha  (Linux/OSX only)

Bulk Importing, Address Books, and ton of bug fixes!  Checkout the "bulkimport" branch to try it out.  "git checkout bulkimport".
Both these new features are a very rough around the edges.  As far as I've tested, they work.  But they need improvement, both in explanation and layout/interface.  I'll get there.  But there's a lot of use cases and corner cases to catch, and I want people to start vetting them immediately.

Code:

sudo apt-get install pyqt4-dev-tools
cd BitcoinArmory
git pull
git checkout bulkimport
cd cppForSwig
make swig
cd ..
python ArmoryQt.py

NOTE: In order to use qrc-resource files responsibly with git, I had to add a step to the Makefile which requires an extra package.  It requires installing the pyqt4-dev-tools package, which is available natively in the Ubuntu repos.  I have updated the webpage with the updated install list (so if you start over from scratch from here, it should work).

This also requires installing a new package if you want it to work on the offline computers, too.  I'll post directions for that on the official release.  I fixed a bunch of offline-tx bugs, but I would wait until this version becomes stable before going through the trouble.

[1714053750]

1714053750

[1714053750]

1714053750

[1714053750]

1714053750

[Foxpup]

It seems to work fine except for crashing in flames if it can't detect my crappy Internet connection:

Code:

Loading Armory Engine:
   Armory Version:       0.72
   PyBtcAddress Version: 1.00
   PyBtcWallet  Version: 1.35
Detected Operating system: Linux
   User home-directory   : /home/foxpup
   Satoshi BTC directory : /home/foxpup/.bitcoin/
   Satoshi blk0001.dat   : /home/foxpup/.bitcoin/blk0001.dat
   Armory home dir       : /home/foxpup/.armory/
/home/foxpup/.armory/ArmorySettings.txt True
Using settings file: /home/foxpup/.armory/ArmorySettings.txt
Loading wallets...
Number of wallets read in: 1
   Wallet (7FVyhG1d):     "Primary Wallet                    "    (No Encryption)
Internet connection is Available:  False
Satoshi Client is Available:       True
Traceback (most recent call last):
  File "ArmoryQt.py", line 1786, in <module>
    form = ArmoryMainWindow(opts=CLI_OPTIONS)
  File "ArmoryQt.py", line 95, in __init__
    self.setupNetworking()
  File "ArmoryQt.py", line 587, in setupNetworking
    dlg = DlgBadConnection(self.internetAvail, self.satoshiAvail, self, self)
  File "/home/foxpup/BitcoinArmory/qtdialogs.py", line 6929, in __init__
    if not haveInternet and not self.main.ignoreblk:
AttributeError: 'ArmoryMainWindow' object has no attribute 'ignoreblk'

So, uh, what's the package I need for it to work offline?

[etotheipi]

Foxpup,

(1) I can see why that line should trigger an error, but I can't seem to trigger it myself.  Bizarre.  Either way I committed and pushed a fix.  Let me know if there's still a problem.
(2) I just added (and tested) a new command-line argument:  "--nettimeout=X" where X is the number of seconds.  I pushed that with the above fix.  Yes, that's one's for you, Foxpup
(3) I already mentioned the package: it's pyqt4-dev-tools.  More-specifically (for other distros) you need to have pyrcc4 in your path.  The "directions" I was referring to in my previous post were for installing pyqt4-dev-tools on an offline computer... i.e. your offline computer doesn't have internet, how do you install it?  (it's pretty easy in Ubuntu, actually, if you know what buttons to press).

A little more elaboration on why #3 was necessary: I previously used pyrcc4 to create a Qt resource file, but that file was 10 MB and ended up in the git repo.  But every time I add a new icon or image, that file is going to change and add 10 MB more to the repo (kind of like the blockchain... it's never going to get smaller).   So instead, I have committed all the individual images, and execute the pyrcc4 call in the Makefile.  Otherwise, the repo might start looking like the blockchain...

[bitlotto]

I'll spend some time thinking about how to enable this responsibly.  Perhaps only in Dev mode...?   (though it doesn't solve the problem of someone sneaking a key into your wallet)

Perhaps you could make the "Ledger"  (Maybe turn into "Wallet Ledger"?) a drop down selection list. Have "Watching" one of the options. Then when "Watching" is selected it just lists all the addresses and the totals for each one. Add a little button "Add address to watch" and the user then inputs the Bitcoin address only. You could just enter any address. It scans the blockchain and finds the total. Something like that perhaps? That way the Ledger and Watching are never together in the same list.

[Foxpup]

(1) I can see why that line should trigger an error, but I can't seem to trigger it myself.  Bizarre.  Either way I committed and pushed a fix.  Let me know if there's still a problem.

Fixed. Don't forget, order matters with the and operator. If the first operand evaluates to false, the second operand won't be evaluated at all, since the and operator is guaranteed to return false regardless of what the second operand is. This is why that line didn't cause an error when the Internet connection was detected successfully: the erroneous expression was never evaluated. The same thing happens with the or operater if the first operand is true.

(2) I just added (and tested) a new command-line argument:  "--nettimeout=X" where X is the number of seconds.  I pushed that with the above fix.  Yes, that's one's for you, Foxpup

Thank you.

(3) I already mentioned the package: it's pyqt4-dev-tools.  More-specifically (for other distros) you need to have pyrcc4 in your path.  The "directions" I was referring to in my previous post were for installing pyqt4-dev-tools on an offline computer... i.e. your offline computer doesn't have internet, how do you install it?  (it's pretty easy in Ubuntu, actually, if you know what buttons to press).

Well I'm guessing (and this is pure speculation on my part) that you'd install it the same way you installed all the other dependencies for Armory? And Armory itself for that matter?

[etotheipi]

(3) I already mentioned the package: it's pyqt4-dev-tools.  More-specifically (for other distros) you need to have pyrcc4 in your path.  The "directions" I was referring to in my previous post were for installing pyqt4-dev-tools on an offline computer... i.e. your offline computer doesn't have internet, how do you install it?  (it's pretty easy in Ubuntu, actually, if you know what buttons to press).

Well I'm guessing (and this is pure speculation on my part) that you'd install it the same way you installed all the other dependencies for Armory? And Armory itself for that matter?

Well, typically you can setup the offline system while online (which takes 3 minutes), and then disconnect it permanently before creating your first offline wallet.  However, this means you may not know how to install new packages, since us Ubuntu folk are so dependent on our versatile package managers.

In Synaptic, after you select all the packages you want to install, you can go to File-->Generate Download Script.  It then creates a script (which is just a bunch of wget commands) that can be run on any linux system to download the package and all of its dependencies.  Save the script to its own directory on a USB key.  Then you just double click from an online computer and downloads everything for you into the same directory.

When you go back to the offline computer with your USB key full of packages, you go into Synaptic again, and File-->Add Downloaded Packages.  You point it to that directory, and it installs them as if it had a connection.  This is a useful feature, because Synaptic knows what dependencies and versions it needs, and will do it The Right Way for you. 

[Foxpup]

Synaptic knows what dependencies and versions it needs, and will do it The Right Way for you. 

Unless one of the dependencies conflicts with your current version of libc6, and APT decides the best way to resolve the situation is to uninstall libc6. The fact that the following command:

Code:

apt-get -f install libstdc++6 g++-4.4 gcc-4.4-base gcc-4.4 libstdc++6-4.4-dev libgmp10 libmpfr4 cpp-4.4 libgcc1 libgcc1 libgfortran3 libgomp1 libquadmath0 libc6-dev libc6 libc-dev-bin make pkg-config libc-bin locales

appears more than once in my .bash_history would seem to indicate that Synaptic does not always know what the Hell it is doing.
</rant>

[etotheipi]

I'll spend some time thinking about how to enable this responsibly.  Perhaps only in Dev mode...?   (though it doesn't solve the problem of someone sneaking a key into your wallet)

Perhaps you could make the "Ledger"  (Maybe turn into "Wallet Ledger"?) a drop down selection list. Have "Watching" one of the options. Then when "Watching" is selected it just lists all the addresses and the totals for each one. Add a little button "Add address to watch" and the user then inputs the Bitcoin address only. You could just enter any address. It scans the blockchain and finds the total. Something like that perhaps? That way the Ledger and Watching are never together in the same list.

It's not a bad idea for helping users do this without shooting themselves in the foot.  But I don't think it solves Red Emerald's problem, which is that he does want this address added to his wallet so that he can then use it for offline transactions.  Implementing your suggestion would be great for letting users simply-watch-and-nothing-else addresses on demand, but doing anything other than monitoring payments/balances would be complicated.

Right now, the only way to achieve Red Emerald's goal is to go to your offline system first, import the key, fork the watching-only wallet, then copy it to your online computer (removing the original).  It's a few extra steps for sure, and it could be improved.

However, regardless of the solution: discussing this just made me realize a very important feature I need to add (and is already kind of in queue to be added):  separate comments file.  Right now, all your address/tx comments are stored in your wallet, and there's no way to merge a new wallet in (as described above) without overwriting all of them. 

I say this is "kind of in queue already", because I will need to do the exact same thing with P2SH scripts:  they will need to be stored in a separate, low-sensitivity file that can be dropboxed or part of some kind of daily backup.  You need that stuff backed up as soon as you create it (comments, not so important, P2SH scripts:  very important), but you don't want your whole wallet file, encrypted or not, sync'd to multiple other servers on the internet. 

[Red Emerald]

I'll spend some time thinking about how to enable this responsibly.  Perhaps only in Dev mode...?   (though it doesn't solve the problem of someone sneaking a key into your wallet)

Perhaps you could make the "Ledger"  (Maybe turn into "Wallet Ledger"?) a drop down selection list. Have "Watching" one of the options. Then when "Watching" is selected it just lists all the addresses and the totals for each one. Add a little button "Add address to watch" and the user then inputs the Bitcoin address only. You could just enter any address. It scans the blockchain and finds the total. Something like that perhaps? That way the Ledger and Watching are never together in the same list.

It's not a bad idea for helping users do this without shooting themselves in the foot.  But I don't think it solves Red Emerald's problem, which is that he does want this address added to his wallet so that he can then use it for offline transactions.  Implementing your suggestion would be great for letting users simply-watch-and-nothing-else addresses on demand, but doing anything other than monitoring payments/balances would be complicated.

Right now, the only way to achieve Red Emerald's goal is to go to your offline system first, import the key, fork the watching-only wallet, then copy it to your online computer (removing the original).  It's a few extra steps for sure, and it could be improved.

However, regardless of the solution: discussing this just made me realize a very important feature I need to add (and is already kind of in queue to be added):  separate comments file.  Right now, all your address/tx comments are stored in your wallet, and there's no way to merge a new wallet in (as described above) without overwriting all of them. 

I say this is "kind of in queue already", because I will need to do the exact same thing with P2SH scripts:  they will need to be stored in a separate, low-sensitivity file that can be dropboxed or part of some kind of daily backup.  You need that stuff backed up as soon as you create it (comments, not so important, P2SH scripts:  very important), but you don't want your whole wallet file, encrypted or not, sync'd to multiple other servers on the internet. 

I don't actually care about offline transactions.  I just want the book keeping.  I've been doing it my way because my way was the only way I could think of.

[etotheipi]

Armory version (0.72-beta)-alpha-RC2   (Still Linux/OSX only)

Still on the "bulkimport" branch!  But stability is rapidly approaching, will merge into master soon.  PLEASE HELP TEST!  (See my previous post about how to run 0.72)

Great News!

• Fixed color schemes!  Looks great with dark desktop themes! (see below)
• Cleaned up address books!  Better user interaction, better information!
• Cleaned up bulk-import!  Warns you about keys already in wallet.

Terrible News!

I have become aware that the Windows version 0.70+/RAM-Reduction is preventing the Satoshi client from writing the blockchain.  This went unnoticed, because you can still send transactions and receive zero-conf transactions, but you won't see any confirmations until you restart Armory!  This is super-top-priority for fixing!

(Linux version works fine, though)

[etotheipi]

This is worthy of an announcement update (so I will post one soon) -- the Windows issues are completely fixed! 

The memory mapping was preventing the Satoshi client from accessing the blockchain file at all!  I hate Windows.  But I'm relieved.  Fwhew!

I also got sick of the -beta-alpha version which made more sense if I was like 2 days from beta.  But I promised a new wallet format for beta, and that might still be a couple weeks off.


Find the new Win64 binary in (version 0.73-alpha):   https://github.com/etotheipi/BitcoinArmory/downloads

[Red Emerald]

This is worthy of an announcement update (so I will post one soon) -- the Windows issues are completely fixed! 

The memory mapping was preventing the Satoshi client from accessing the blockchain file at all!  I hate Windows.  But I'm relieved.  Fwhew!

I also got sick of the -beta-alpha version which made more sense if I was like 2 days from beta.  But I promised a new wallet format for beta, and that might still be a couple weeks off.


Find the new Win64 binary in (version 0.73-alpha):   https://github.com/etotheipi/BitcoinArmory/downloads

Did you mean "The memory mapping wasn't preventing the Satoshi client from accessing the blockchain file at all!"

[etotheipi]

The memory mapping was preventing the Satoshi client from accessing the blockchain file at all! 

Did you mean "The memory mapping wasn't preventing the Satoshi client from accessing the blockchain file at all!"

No, I said it right.  In Windows, my memory-mapping was locking out all other processes from the blockchain file.  This meant that the Satoshi client couldn't read or write the blockchain file at all -- no validation no block updates.  Whatever block you were on when you started Armory, is where you were stuck (though incoming or outgoing 0-conf tx worked fine, they were just stuck 0-conf).  But I figured out the correct flags for the mapping in Windows, and it should all work now!

[etotheipi]

Just posted version 0.74 which fixes a few bugs in 0.73:

-- Multiple-instance prevention didn't work in offline mode.  It does now.
-- Importing addresses and removing imported addresses now updates address list immediately and correctly.  If you don't feel like upgrade, be sure to close the wallet props window and reopen after an import.  The import works, but it doesn't update the table.

This wasn't a big enough issue to be worth updating the announcements thread.  But the import/remove thing was bugging me, so maybe it was bugging you, too.

[ThiagoCMC]

Using Armory 0.7 (and Bitcoin 0.6.0) for Linux on a daily basis... Thanks!!

[istar]

As soon as Armory support comps with less than 4 gig ram. I will start using it.

[etotheipi]

As soon as Armory support comps with less than 4 gig ram. I will start using it.

istar,

It has supported them since version 0.70.  Windows systems with 1.5 GB+  and Linux with 0.5 GB or more.  Windows didn't react so nicely to my RAM-reduction upgrades, but it does work, and it shouldn't kill the system (it looks like it's using a lot of RAM, but it's mostly virtual which means that it can/will give up some when other programs are loaded).

The only restriction is that it won't work on 32-bit Windows.  32-bit Linux is fine.

[kr105]

I was going to improve your Windows mmap implementation, but i just read the README and i'm too lazy to set-up the build enviroment on Windows right now (3 am). So i'm going to sleep a bit and then help!

[etotheipi]

I was going to improve your Windows mmap implementation, but i just read the README and i'm too lazy to set-up the build enviroment on Windows right now (3 am). So i'm going to sleep a bit and then help!

This sounds fantastic!  Admittedly, my windows MMAP/MVOF skills are below par.  I couldn't understand why declaring sequential scanning didn't improve the load time at all, and why it still seems "finnicky."  Please PM me with the information and I can put the improvements directly into my branch, rather than dealing with getting the build environment setup. 

If it's more involved than just changing flags... well the Windows build instructions on my website are really good!  (it's not pleasant, but I've documented it thoroughly enough that it worked for a lot of people building it).  Just be aware, that it didn't go smoothly last time I tried to build it in MSVS 2005.  I use 2008 and I've heard 2010 works, too.

[kr105]

I was going to improve your Windows mmap implementation, but i just read the README and i'm too lazy to set-up the build enviroment on Windows right now (3 am). So i'm going to sleep a bit and then help!

This sounds fantastic!  Admittedly, my windows MMAP/MVOF skills are below par.  I couldn't understand why declaring sequential scanning didn't improve the load time at all, and why it still seems "finnicky."  Please PM me with the information and I can put the improvements directly into my branch, rather than dealing with getting the build environment setup. 

If it's more involved than just changing flags... well the Windows build instructions on my website are really good!  (it's not pleasant, but I've documented it thoroughly enough that it worked for a lot of people building it).  Just be aware, that it didn't go smoothly last time I tried to build it in MSVS 2005.  I use 2008 and I've heard 2010 works, too.

Thanks, i'm going to build it and post my results (I'm using VS2010)