Merged mining + timetravel fix @19200 - must upgrade

[moa]

Needs more testing and network is too slow to test in anger. We should have enough time to gather data for a few days after next mining boost (20,160) and do any possible mods before network update having to collide with the other 24,000 changeover mods.

So say 21,000.

[jtimon]

Don't know what block number would be more appropiate, but again the sooner the better.
I guess most namecoin miners follow this forum, so I wouldn't be very scared about many people not changing their miner. When is tested it can go to production.
I don't mine, but I can help "spreading the word" in the bitcoin forum. Not only for namecoin miners in there, but for those who only mine for bitcoin.
This is very exciting.

[thirdlight]

Block 21,000 will be right in the middle of the next surge - which seems very risky to me.

If we're not going to go for a block before 20160, we should definitely go for one after 22176.

I like 20000 for before or 22300 (22176+120+a bit) for after.

[moa]

Block 21,000 will be right in the middle of the next surge - which seems very risky to me.

Risky for who?

When merged mining kicks in these 'surges' are going to look like a picnic. It will one 4xdifficulty change after another as the full hashpower of btc miners is turned toward the nmc network.

If you do it at 22,176 those 2016 blocks after it gets turned on will be gone in 2 days or less. Not much time to make network wide changes before the previous validity mod takes effect, potentially complicating a hectic situation.

Maybe leave it until 25,000.

[moa]

Is there a detailed report anywhere?

The report is basically, checkout the current git master. Build and run it with a clear datadir. It doesn't download any blocks or make any connections. I haven't tested your latest changes.

Thank you, I found the issue. A new version tagged nc0.3.24.53-rc2 is up on my repository. The previous version rejected current production blocks because it was checking for the new block version ID prematurely.

I gotta ask vinced, because no-one else seems willing to.

Are you yourself even test running the code once against the network before you put it up on git master?

[thirdlight]

Risky for who?

Just that blocks will be being found very quickly. At 22176, even if all bitcoin miners stay with namecoin, it will be 4x slower, then another 4x ....

My guess is they'll jump off again. Then they'll come back as pools implement support for merged. Probably smaller pools first, then the larger ones when there's evidence of stability. No more surges, just increasing hashing power. A bit like when gpu mining caught on.

[smoothie]

My vote is for merged mining to take place before the block 24,000 changes. Also I believe the sooner the better given it has been tested.

[nodemaster]

Latest Git (should be RC2) is segfaulting if I try to run namecoind in testnet mode. Starting on production net seem to start without a problem. However I get no connections. Bitcoind (same repository) is running fine on testnet. Am I doing something wrong or is there anybody else having this issue?

Edit:
On production it seems to be the same issue doublec had with RC1. But I also downloaded RC2 Tarball to make sure, I have the right codebase.

[iopq]

oh my god, it will be october before the next mining party, and even THEN we won't be at block 24000

is it impossible to change the difficulty rules?

[vinced]

Anybody in favor of moving up the starting block number to 19500 if no bugs are found in the current version over the next week? At the current block rate, this would be about four weeks.

My concern with this is it not being enough time to get people to update the client. Ideally we'd have it set to a block in the future and have a couple of releases before then to enable people to update. Then by the time it activates everyone is on the new version. Are you able to send a network message to clients asking them to upgrade?

I did not change the "emergency broadcast" key, and only Gavin has the key. That was an oversight. I will change it for the next release.

I think most people would check the forums once they notice that mining has slowed down by a factor of 3. This is assuming 66% of mining power is on the new version. Maybe we could count the pool mining power that is represented on this thread and see what % it is?

I think we should aim to do the switch during a slow period, so that a 5x slowdown will be very obvious to the 1/3 that did not upgrade.