Calculating a block - merkle root question

[theymos]

You need to modify the Bitcoin code for that. You can't do it just by using getwork.

[Pibers]

Hey,
I took data with getwork while block 129593 was the newest on blockexplorer.com and also with bitcoind getinfo it said "block" : 129593.
But now I have a Problem. This is what "bitcoind getwork" gave me (already formated):

Code:

00000001
9ba7ad42f0b7c0c3bcec91db32107ad5ffaa4e07af2ed372000006e400000000
c2aa00b9a62ed9f4642c35b12cea291fc778be2e7b29a3879ffaf9d3c567cdd9
4df0d584
1a1d932f
00000000
000000800000000000000000000000000000000000000000000000000000000000000000000000000000000080020000

So the previous hash (if I reverse it) is:
000000004e600000273de2fa70e4aaff5da70123bd19cecb3c0c7b0f24da7ab9

But bitexplorer said that previous hash is:
000000000000169df290bd7628e59dae18405aba2f660b4ff832f93dc9a10fb0

The first hash also hasn't enough zeros I guess!? What's wrong there?

What can I do to convert the data from blockexplorer to a valid data string for getwork? I would like to test to send a valid block (after it is solved) and see no other way to do it than taking the data from blockexplorer...

[theymos]

The hashes have been reversed on 8-byte boundaries. I guess this is a hashing optimization. I'm not sure.

[Pibers]

Mh thanks but I don't get that but I think it is not so important
Though I have one last question. When I want to calculate a block am I on the right way with that?
- I take the data string from getwork (without line breakes)
- I try many values at the place in the string where the nonce is (should be from char at position 153 to 161)
- After each nonce value I SHA256 the whole string
- Now I test if there is the right count of leading zeros!?

I dont understand the last step. Afaik I have to check if the hash is smaller than the current target. But how can I do this?
For example the hash from previous post:
000000000000169df290bd7628e59dae18405aba2f660b4ff832f93dc9a10fb0

How can I check if it is valid? (I dont mean the check bitcoin does like if the merkle tree or the timestamp is right. I mean just the check if it hast the right difficulty)

[theymos]

They're both large numbers. Just see if the hash is less than the target.

You shouldn't do this every try, though. It's more efficient to count the zeroes and then do a full test when you have a candidate.

[Pibers]

Oh yes, now where you have said it it is obvious  
Thanks!

I guess you have no solved old data block or know where I can get one? Maybe for testnet?

[theymos]

I don't have old getdata output. Here's a block header you can try hashing, though this does not have the reversal on 8-bytes:

Code:

010000001d8f4ec0443e1f19f305e488c1085c95de7cc3fd25e0d2c5bb5d0000000000009762547903d36881a86751f3f5049e23050113f779735ef82734ebf0b4450081d8c8c84db3936a1a334b035b