Fix from luke-jr, I reviewed the code but have not tested.
Rough sketch of a test plan:
(extraNonce being reset due to just-occurred time change after nTime is set)
First of all, can we get commitmsgs that are readable?
Ok I did the steps above in a fresh build of 0.4.0rc1 with the testnet-in-a-box files, and landed at code:-12 error: Keypool ran out, please call keypoolrefill first
I then restarted my daemons with the -gen tag. Up to about 20 generated blocks now.
My Ubuntu skills are lacking, the txids look unique to me; but obviously that's not good enough. Any help with http://bitcoin.stackexchange.com/questions/706/how-do-i-sort-the-output-of-listtransactions would be appreciated.
Lastly, I'm not familiar with RPC - is there any resource for how I would do a getwork call wtih RPC?
getwork is how external miners ask for block headers to work on, so to test it you'd fire up a miner and tell it to ask for work from your bitcoind.
There's an external python-based miner in the config/pyminer/ directory (named pyminer.py) that aught to be handy for testing (it is very slow, though). If you're running testnet-in-a-box, note that you'll need to be running two bitcoinds talking to each other for any mining to get done.
I just tried:
pyminer.py bcc1.cfg
... and got: 2 mining threads started Wed Sep 7 19:08:12 2011 Miner Starts - 127.0.0.1:9000 HashMeter(0): 1000000 hashes, 249.67 Khash/sec HashMeter(1): 1000000 hashes, 255.08 Khash/sec
where bcc1.cfg contains settings that match my weird testnet-in-a-box setup (running with rpcport=9000):
host=127.0.0.1 port=9000 rpcuser=abc rpcpass=123 threads=2 scantime=60 hashmeter=1
The commits are a bit messy - lots of changes that are not used in later commits. ACK on the actual code change though (though it needs testing).
Tested, no duplicate txids.