Topic: Calculating a block - merkle root question - page 2. (Read 4783 times)

"midstate" and "hash1" are optionally used for a mining optimization. Some of the work doesn't need to be repeated for each attempt: this is the data that allows you to skip it.

"data" is what you're hashing. You hash this, modify the nonce in the data if the hash is not below the target, and repeat.

You only need to hash the header, which is the "data" getwork gives you. The block body is not hashed by miners. Bitcoin deals with this.

When you return a block, Bitcoin just checks to see if it's valid. It doesn't need to match anything.

I assume that you pass hexadecimal data back to getblock. I'm not sure.

Thank you again, you're great help
I found out how to "get work" and I also found the format description on that site: https://en.bitcoin.it/wiki/Original_Bitcoin_client/API_calls_list
What I understand ist "target" but I have no idea what I have to do with midstate, data and hash1?
Where is the body with the transactions? How can I calculate the block without that?
Maybe bitcoin don't give me this data but only what I need for calculation? But when I calculate a while how did bitcoin now with which data I made my calculation when I send it back?
And thats my next question. I can send data back with "getwork [data]" but in which format?
Sorry if that's explained somewhere in the wiki but I haven't found something (nor on google).

It's an official RPC method now. Every recent version of Bitcoin has it. It allows miners to get the necessary data from Bitcoin.


"Keep doing this until you have only one hash left." If you have only one transaction in the block, then the hash of that transaction is the Merkle root.


No. Only transaction fees.


That could happen, though it is unlikely. The ties will be broken eventually. Whoever gets the last block in the series of ties will decide it.

Thank you so far, that helps
A few questions are left...

getwork is new for me! I googled it and wonder if you mean m0mchils patch? I patch the bitcoin client (or server?) with it and so I'm able to get block data and return it to the network if I find a block? That would be easier than I thought but not really a own miner implementation I guess?

If I understand you right I could take only one of the new transactions and double SHA256 it with the concatenation of itself? Like SHA256(SHA256(hash + hash))?
So what I have then is a really small merkle root isn't it? I could use it for a valid block but I guess that isn't really good for the quality of the network!?
Is there nothing to ensure that as many transactions as possible will be putted in a new block?

Only the longest chain is valid right? So there is a tie of two chains cause of two valid blocks and the next block for one of it decided that this on is the valid chain. But now there could be also been found a new
block for the other chain and after that one again one more new block. So now the other chain is longer than the first one! Is this one now the new valid?? My englisch is too bad I hope you understand

Thanks!

Miners don't deal with that. Bitcoin does that stuff and returns the result to miners in getwork.

Here's an example of a tree:
https://en.bitcoin.it/wiki/Dump_format#CBlock
If you find the full hashes used here (search Bitcoin Block Explorer), you can try hashing them yourself. They will work.

You take each pair of transaction hashes, concatenate them, and hash them twice with SHA-256. Keep doing this until you have only one hash left. When there is an odd number of hashes, concatenate the last hash with itself.

So your example is correct if you double each hash.

The top hash (Merkle root) is included in the block header. The transactions are in the block body. Intermediate hashes in the tree are not included in the block.

Bitcoin has a database of all transactions, which lets it know which transactions are new. It also needs to check the validity of all transactions before including them.

In the case of a tie, only one block will end up being accepted. Whoever ends up solving the next block will decide which one.

Bitcoin gets transactions from the network. You're not forced to include any transactions, though in the future most of your income will come from transaction fees. So your block is still valid if you miss a transaction (or even all transactions).

Again, nothing I described in this post is done by mining software.

it would be also helpful to understand how a miner chooses or is forced to choose amongst tx's?  why not just choose the one simplest tx and hash that to speed block win?

Hey,
I've read a lot but there still are questions for writing my own (test-)miner
I guess I know how to get all data for calculation a block except for how to 'calculate' the merkle root.
I far as I understand I first have to get all recent transactions. Let's say these are only four with the hashes 1, 2, 3 and 4 (I know that this aren't real hashes).
Now I take these and hash them like this:
Is that right? And how exately do I hash them? Is it
where + is writing it just one after the other?

Do I have to write just the masterhash in the block or all transactions? (On blockexplorer.com there are all transactions but I don't know...)
And how do I know what are new transactions and what are old? I guess I look in the old block but what is the newest transaction in the old block?

If I now have the (for example) four newest block and the right merkle root what happens if a new transaction is made while I calculate the block? Is the block still valid?

It woul be so nice if someone could explain me that. Thank you!

Oh and one other question.. what happens if two users calculate a valid block at the exact same time? I guess both blocks could be share through the network and at the and one half has the one and the other half the other block. Which one is the valid then!? Thanks!