Cannot send from Armory

[http]

Hi, I'm using Armory 0.93.3 with Core 0.11.1 and everything seems fine. Armory client is in status Connected and wallet shows correct amount.
Now I wanted to make a purchase, but when I click the actual "Send!" button, nothing happens. No error, dialog remains open. No asking for unlocking the wallet.
When I leave the default fee of 0.0001 and click the send button, I get a dialog that the fee is too small and should be increased. If I click OK, the fee gets increased in the dialog, but also nothing else happens, not even when I click "Send!" again.
I've seen that Armory 0.95.1 with Segwit is out (no idea what that is), but shouldn't be sending work as expected with the 0.93.3 client? Is there a known bug?
In the log there is no error or anything I could see. Is this a known issue? I couldn't find anything. I'm using Win10 (anniversary update) in 64 bit and the wallet was created with an older Armory version.
I thought of changing the wallet password to see if unlocking works, but I'm afraid now to change anything if there is something wrong anyway.

[1713498598]

1713498598

[1713498598]

1713498598

[1713498598]

1713498598

[goatpig]

Hi, I'm using Armory 0.93.3 with Core 0.11.1

That's an old stack. You should consider upgrading.

I've seen that Armory 0.95.1 with Segwit is out (no idea what that is), but shouldn't be sending work as expected with the 0.93.3 client? Is there a known bug?

I'm not sure what the issue is, but I know that it wouldn't happen in 0.95. At any rate, try manually setting a higher fee to spend, looks like the automated fee bump is failing.

In the log there is no error or anything I could see.

That's really for me to say. Post your logs.

I thought of changing the wallet password to see if unlocking works, but I'm afraid now to change anything if there is something wrong anyway.

You did not get as far as unlocking your wallet.

[http]

That's an old stack. You should consider upgrading.

Yes. I know. I just came back without doing much with Bitcoin for at least one year and find my environment somehow broken although there was no blockchain fork as far as I know. First I'd like to find out what is wrong before considering any changes. Today I found out about the Armory company issue and that you took over the client maintenance. Thanks a lot for that! Let me know where to donate; I'll send something as soon as sending works again for me.  
I won't tell here what amount I have in my wallet (maybe just a few Satoshis, maybe millions), but you have to always assume that people fear of losing money (and there could be millions in the wallet). Therefore any change can potentially break things or make it unsafe. I didn't know that Armory was taken over and how trustworthy the new "organization" is and all that. Anyway, I'll probably upgrade soon. Forget this side topic.

I'm not sure what the issue is, but I know that it wouldn't happen in 0.95. At any rate, try manually setting a higher fee to spend, looks like the automated fee bump is failing.

You can't ever be sure that it wouldn't happen in 0.95 before finding the reason. I don't think it's fee-related, because adjusting the fees is working. The amount just doesn't get sent out. Anyway, I tried today again and also with setting a higher fee. it's a relatively small transaction (only one input address, two output addresses) with an amount 0.024579 and fee 0.0007 (more than enough from what I read) and it still failed without error message.

That's really for me to say. Post your logs.

You're right here of course. And I was also wrong saying that there is nothing; I looked at the wrong log (the cpp log). In the normal log you can see Armory crashing.
And when I was looking at the cpp log, there was not date/time info, so it was difficult to find out what was new, but I searched for "error" and couldn't find anything interesting.
Here's the log:
http://pastebin.com/raw/rH2mbmyZ
It contains both log and cpplog, but only the entries after Armory has started up.
I clicked a link somewhere else (Bitcoin payment link), therefore you see the entries "already open" and that stuff.
I censored the wallet name, my Windows user name in the directory names and the exact bitcoin: link that I clicked, as the same error also happens when I copy/paste the destination address manually into the send dialog.
So what sticks out is the crash:

Code:

2016-12-28 21:42 (INFO) -- TxFrames.pyc:720 - Change address behavior: NewAddr
2016-12-28 21:42 (ERROR) -- Traceback (most recent call last):
  File "ui\TxFrames.pyc", line 785, in createTxAndBroadcast
  File "ui\TxFrames.pyc", line 766, in validateInputsGetUSTX
  File "armoryengine\Transaction.pyc", line 2148, in createFromTxOutSelection
  File "armoryengine\Transaction.pyc", line 2058, in createFromPyTx
  File "CppBlockUtils.pyc", line 2014, in getTxByHash
RuntimeError

Any ideas what is causing this? Is there a way I can debug this code? I'm not familiar with Python, but I know C++ very well.
Where did it crash exactly? In CppBlockUtils.pyc/2014ff? Where can I find this old file? I can't find it in the etotheipi guthub repo.

[goatpig]

I won't tell here what amount I have in my wallet (maybe just a few Satoshis, maybe millions), but you have to always assume that people fear of losing money (and there could be millions in the wallet). Therefore any change can potentially break things or make it unsafe. I didn't know that Armory was taken over and how trustworthy the new "organization" is and all that. Anyway, I'll probably upgrade soon. Forget this side topic.

Of course you should always tread with caution. For what it's worth, I'll lay my case again for taking over development. I'm a former ATI employee. I was in charge of basically most of the open source development since I started working on Armory (around 0.91), with the exception of actual crypto features which Alan handled directly.

The last FOSS ATI version (0.93.x) is over 90% my code. In other words, you are already running a lot of my code.

I've also made a point of leaving all wallet critical code untouched up until and including version 0.95.1 (which can parse SW blocks). The next version will have fresh wallet code. If you want to keep my work in check, this is the version you need review. I'll be posting the particular files for people to focus their reviewing on when the code goes into testing.

However, being conscious of the needs of the more paranoid, that new code will only operate when signing SegWit transactions. The old code is still there and active when using Armory with its legacy features.

You can't ever be sure that it wouldn't happen in 0.95 before finding the reason.

I know because there can only be 2 types of errors in your case, one being fees, the other being something I fixed some 10 months ago for 0.94. 0.95 also has fresh, proper P2P code to push transactions and handles fees in a much better fashion.

I don't think it's fee-related, because adjusting the fees is working.

It's not indeed. A rebuild & rescan should do the trick I believe.

Any ideas what is causing this?

I forgot what it was. You can crawl through the commit history for 0.94, it should be pretty obvious then.

Is there a way I can debug this code?

Yes but you need to know your way around either GDB or MSVC.

Where did it crash exactly? In CppBlockUtils.pyc/2014ff? Where can I find this old file? I can't find it in the etotheipi guthub repo.

That file is the result of the SWIG compilation, which bridges C++ code over to Python. What you are looking for is the actual code, which you can find here:

https://github.com/goatpig/BitcoinArmory/blob/v0.93.3/cppForSwig/BlockDataViewer.cpp#L267

At any rate, you should consider upgrading for all the benefits the new versions provide. You won't have a choice if you want SegWit or to even run a SW activated node, as anything prior to 0.95 will choke on SW blocks.

[http]

Thanks for explaining more details! I'll definitely update my client sooner or later.

A rebuild & rescan should do the trick I believe.

Indeed. This fixed it. Silly me. I thought this was to rebuild the blockchain, which could take days.

Yes but you need to know your way around either GDB or MSVC.

I'm familiar with debugging in general and also with MSVC.

That file is the result of the SWIG compilation, which bridges C++ code over to Python.

Yeah, I found (after my last post) that it was a library, but as I don't know Python I still didn't figure out the connection.
Maybe I'll try to build the Windows client myself as a tiny first step. Are the build instructions still valid or did you update them somewhere?
On the other hand, I'm pretty busy as usual (it's now 2.30am and I have to go to work tomorrow), so maybe I never get anything done.
Looks like you don't want to talk about donations. If you want, put an address into your profile.
Thanks again, great work!

[goatpig]

Are the build instructions still valid or did you update them somewhere?

https://btcarmory.com/docs/building/

These were valid last time I checked.

Looks like you don't want to talk about donations

I do not accept donations atm.