Ah. This makes again much more sense! Of course. Guess  I will have to read some more satoshi c++...

Just curious: How is the longest chain picked if there are two chains with an equal amount of proof-of-work?

  Looks like we are at the same stage of understanding bitcoin 

This sunday I am working on exactly the same issue. I found three aspects helpful:

1) The decription at https://en.bitcoin.it/wiki/Block_hashing_algorithm gives a rough idea

2) Reading the source of the satoshi implementation is...um...a good exercise in c++ fluency. However it is helpful for this question. The issue is: You have to get all the endian-nesses and all the sequences right. Satoshi implements an elaborate serialization concept and derives the has from there.

3) Probably the most helpful thing in this regards is the pastebin reference from (1) http://pastebin.com/n8UEGA86 It contains a c implementation of exactly the part you are looking into. I did not try it myself (but probably will do so eventually).

That said, let me make a general recommendation: Read the Satoshi implementation. It is not well documented, has a file structuring which is quite hard to read, uses boost library calls and has all kinds of twists and turns. However, it does pay off. Satoshi has numerous twists and turns and ideas and it takes a while to understand those. But it is not only every elegant (you can learn something about programming...)  -  it also contributes to the security and stability of the system.

I would recommend against php based implementations. And I would recommend against doing own implementations in bitcoin before you have completely digested the concepts in the Satoshi implementation. It took me 6 weeks of debug instrumentation, code commenting, code analysis and more - and as a result of this I ended up with this opinion.

Good luck.

Yup. Of course.

I did not phrase the question clearly enough. I was thinking of the following situation: A node is booted, bitcoind is started...reads the disc files and finds a block index where there are TWO longest chains. The program will prefer one of them. Which one will it prefer? I was not able to find that answer. I now realize, that it is in fact not really important. And I realize that I will have to spend a day or two understanding block reorganization... :-)

Actually, the terminology of longest chain is ill defined. There can be several chains of maximal length.

And again: If this thing (bitcoind) should eventually takle over management of my money - I want to make sure, as computer scientist, I understand every single bit of what it does. 

Thank you, Joel, for your help ! It clarified the issue.

Trying to work my way through the code  according to Richard Feynman's principle of learning "What I can't produce - I do not understand".   

So, I reached another place I do not understand. 

Every node always considers the longest chain the correct one. Yes?

Ok. What happens if there are two longest chains.

Well, this is not a problem for the future of the entire system since, eventually, there will be a longest chain again. Odds are exponentially decreasing for a "two equally long longest chains" to survive in the long run.

That said: A single node (and, similarly, blockexplorer) has at any given moment in time a concept of what is the longest chain (even if there are two equally long chains). I checked the code. In the -printblock functionality there is no clear answer to this and I did not yet manage to digest the chain reorganization algorithms to understand this aspect.

So, assume there is this situation:

               B-C
              /
GENESIS-A
              \
               X-Y

What does the individual node or the block explorer consider the longest chain ? GENESIS-A-B-C or GENESIS-A-X-Y ?

This question might be considered academic, since in the long run one of the chains will stabilize and become the longest.

But there might be one interesting aspect here. I think, for quick chain stabilization, it is important that the nodes pick the same chain as the correct chain. Assume, 1/2 of the nodes pick GENESIS-A-B-C and the other 1/2 of the nodes pick GENESIS-A-X-Y. In this case the "battle" for the right correct chain might take much longer. Now assume that all nodes pick GENESIS-A-B-C. In this case the "battle" will be over quite quickly (not necessarily in favour of GENESIS-A-B-C though, but the chances of oscillating back and forth should be smaller).

Any ideas on this?

That's great and I'll check it. As a general idea: The more open this is, the easier it is to cooperate...


Governments have been using the argument of child molestation and terrorism for more than one decade to build up a more than Orwellian surveillance infrastructure. Our luck is that - until now - it has not really been used to the full extent of its technical possibilities. Governments are already using this line of reasoning against bitcoin and we will see more intense attacks in the future. There is no advantage for political or financial institutions in bitcoin so they will use every possible approach to destroy it, once it becomes a factor of true public importance. There is a reason why Satoshi is a pseudonym.

Sounds great. If I didn't already have a job which presumably is more stable than a bitcoin job I would apply :-)

Still, would be interested somewhat in teaming / cooperating, if you have a model for this.


No. I clearly and certainly do not want governments in bitcoin. I want areas which are free from regulation and free from governments. The p2p nature of bitcoin makes it possible to have a working system without governments. And even more: For many of us the perspective of working towards a government-free and regulation-free area is the very reason of being active in the area.

I do not mind that my two 6990s make a lot of noise, heat my room and comsume a lot of power. I never check mining profitability and BTC value versus power consumption. I invest a lot of time and effort into programming (among others, I have a version for producing my own genesis block, I have my own block explorer, and more, and will be forthcoming with this on github.).

But the moment, governments touch bitcoin, I will pull the plug. Or, rather, have my bitcoin client run on a different config file.

To me, chatting with my neighbor or sending a few encrypted bits and bytes to my network friends, is a core right I exercise without government intervention. Bitcoin, to me, is the extension of this right into the monetary area. Income tax on bitcoin is fine (as part of tax law). But no policy making in bitcoin with the exception of the individual user making his/her individual decisions in front of their computer screens.



 

Really? Gr8 !!

@Joel: Thanx !

Zwei Ideen.

Statt Dummy Plug klappt bei mir ein älterer KVM Switch zur Bildschirmumschaltung, den ich noch rumliegen hatte.

Einige Mining Programme haben einen Flag / Switch, mit der man den Miner bitten kann, hie und da ein paar "idle " Millisekunden einzubauen - die steht die GPU dann für das office Paket zur Verfügung. Praktische erfahrung damit habe ich aber nicht, weil ich selber auf einem mining only rig mine. (den desktop brauche ich, wenn mir der wgen hitze oder was auch immer abschmiert habe ich troubles mit meinem job). Erfahrung kann ich aber bestätigen, die paar Stündchen, die ich am desktop gemint hatte, mit Gtx 285, war das bildschirm handling unter aller s** und auch der 6990er mining rig blinzelt erst mal verschlafen, wenn ich auf seinem screen was will.

+1

I am trying to make my daily advancement in Bitcoin knowledge and got stuck again.   

Trying to understand the blockchain a bit better, I stumbled upon this transaction here:

http://blockexplorer.com/tx/719eb1028a473ac8e804be26542ba3a90effd81e852f83774042ece9ff5bdd04

Looks like this is 1AZUP.... spending some of his/her hard earned bitcoins. But there are three things I do not understand.

Where is the economic sense of paying to 113 recipients in one transaction?

And how would one generate such a transaction?

I could come up with an idea to these two aspects: Probably the tx has not been generated by the GUI client but by a command line or RPC client. And, maybe, it is a mining pool paying out all the fellow miners in a single efficient transaction. Smart.

But how likely would it be that the recipients would ALL have their addresses start with the same prefix of "1Kr"     

This probability is ...um... zero. 

So this is not the correct interpretation.

So what is the correct interpretation here ? What is the deeper mystery of things going on here

Thank you very much for the hint ! 

I checked the temperatur and it looks to me that for some reason the temperature with the new Catalyst is higher (maybe they use the GPU more efficiently). I had had some freezes earlier when the temperature went over 90 celsius - so I reduced the frequency to stay around 83 at most and checked during the night. Rock stable until now. So currently I suspect it was a different reason...

Nope. We need a fix for this. It happens to often. "You can..." at the present stage of the project is not very encouraging for new adopters.

It does not seem to be a noobie error. I can reproduce the error on three different windows machines. Since I am hacking my own variant of the reference client on Linux I did not care too much about it but with the activity this thread gets here we need a quick fix. I do not have a working bitcoin compliation environment for windows at present, so I cannot check the reason of this or make a patch myself - but I think the matter is urgent.

Never change a running system. 

I felt tempted to upgrade AMD Catalyst from 11.5 to 11.6 (link at AMD). Afterwards GUI miner did not work so I removed and reinstalled the new version.

This is one windows 7 / 64 bit.

Since then my system is unstable. 

This means: I get a frozen system in 10 - 20 % of the restarts and the system stops mining and freezes after some 8 - 15 hours of mining.

No error logs or events pointing out the reason.

I am not overclocking or otherwise provoking things and before that on 11.5 and the May (?) version of GUI miner everything was rock solid and stable.

Any similar experiences ? Any ideas ?

Bei mir laufen 2 6990 luftgekühlt. Eine 6990 wassergekühlt hat nur 10 sekunden durchgehal...schmor. :-( Soll ein Defekt der Serie sein.

Ich habe die luftgekühlten 6990 in einem offenen Gehäuse, Lüfter der GPU auf 95% in der Software festgeklemmt, Frequenz ein wenig runtergeregelt so, dass ich typischerweise 80 - 88 Grad Celsius habe. Tut eigentlich ganz gut, seit 1 Monat recht stabil. Plane Aufrüstung auf dritte 6990.

Those that are designated "strange" in block explorer. 

It looks like I didn't succeed in bringing home my message.

The issue is: Bitcoin is still at the very beginning. While I am very smpathetic with people who lost money due to a stolen wallet or a crash at Mt Gox or some other reason - I am a bit upset at those who just consume and, if it does not work exactly as expected, whine and complain.

Why don't you open up a competing exchange, if you do not like Mt Gox ?

Why don't you provide some software fix to improve wallet security ?

Why don't you write up a guid on bitcoin ?

If you want this endeavour to succeed - we all must contribute.

Some do, and some do very much. Thanx to the supporters of bitcoin !

I am following this Mt Gox thing on and off, once in a while.

I was asking myself how many of those who complain about their service ever had the idea (courage, ability, balls) to offer a similar service and face the possibly conneted riscs.

Disclaimer: I have no account at Mt Gox (until now) and did not lose money with them (until now).

I had mine burn up after 30 seconds. Looks like there was a problem withg the cooling block. Dealer told me he coul dnot replace it since they discontinue the watercooled 6990 by PowerColor as they have way too much returns with this card.

That's the point. If.

Currently, it looks like, there are three types of transactions. People using the satoshi client, special institutions like miners, Mt Gox, exchanges and stuff - and a rest consisting of "strange" transactions directly fed into the system.

I am not interested in what could be done with the current system but what IS done with it. I have not yet reached all conclusions, but you can see quite a bit more in network analysis than I feel comfortable with. That's why I am concerned.

I hope, if I have a bit of time and in a few weeks, I can be more specific.