search

dogecoin / dogecoin

very currency
MIT License
14.43k stars 2.85k forks source link

The "making 1.21-dev great again" plan #3349

Open patricklodder opened 11 months ago

patricklodder commented 11 months ago

Currently there are several issues with the 1.21-dev branch that block the CI and other testing efforts:

  1. The macOs Native CI job needs to be converted to arm64 because Cirrus no longer supports Intel Mac native builds - reported here
  2. Newer boost versions cause several issues on x86_64 linux and ARM macOS reported here and here
  3. The crc32c subtree needs updating to support non-linux ARM, reported here
  4. Porting #3308 will fix bdb5 inclusion for mac in the same way it did for 1.14.7-dev
  5. secp256k1 subtree needs an update because that has evolved to be more stable now, and I've seen the occasional unit test error in the wild on the current bitcoin-core/secp256k1@3967d96 we have
  6. contrib/devtools/security-check.py does not support arm64-darwin, reported here

Because at least 1, 2 and 3 needs to be done to get a working macOS native CI, I think we should just disable that CI job for the initial steps of...

... the plan

Please let me know if there are any objections to this plan.

patricklodder commented 11 months ago

Additional finding from #3260 that I forgot to mention: contrib/devtools/security-check.py does not support arm64-darwin at this time, so we will incur some tech debt until we port in fixes from bitcoin/bitcoin. I'll add "updating devtools checks to support arm64" as an additional item under 4.

JunsenD commented 11 months ago

Hey Pat, thanks for the contribution. I am just wondering is possible dogecoin could have the same feature EIP-4844 (sharding) as TPS enhancement in the future plan?

patricklodder commented 11 months ago

Not in that exact form (we have Bitcoin script, not EVM) but you can probably hack together optimistic rollups today (but I wouldn't recommend it, it will be very hard to do securely.) However, if the Dogecoin network wants to adopt taproot at some point, things like TapRet and BitVM will become possible, permissionlessly. And people have only just started to look at the potential application of rollups using taproot - it's an interesting space to watch right now.

JunsenD commented 11 months ago

Not in that exact form (we have Bitcoin script, not EVM) but you can probably hack together optimistic rollups today (but I wouldn't recommend it, it will be very hard to do securely.) However, if the Dogecoin network wants to adopt taproot at some point, things like TapRet and BitVM will become possible, permissionlessly. And people have only just started to look at the potential application of rollups using taproot - it's an interesting space to watch right now.

BitVM is an interesting topic and dogecoin networks would explode if that being implement. I am not sure what X is doing right now, ppl discuss a lot X will onboard doge later. In your opinion, how does dogecoin network react later?

cedricfung commented 10 months ago

What are the main goals of 1.21? Are we going to have SegWit addresses and CSV op code? I guess no taproot yet.

patricklodder commented 10 months ago

What are the main goals of 1.21?

The goal is updating the codebase safely to a point where we can propose protocol improvements without being scared of unforeseen consequences, like happened with 1.14's CLTV softfork. Basically, creating a baseline from which we can work linearly from there on out.

Are we going to have SegWit addresses and CSV op code?

1.21.0 should prepare for at least CSV and hopefully segwit in a slightly modified/optimized form, but do note that in earlier discussions, devs agreed that actually proposing activation for these should be done in a follow-up release, making sure that we have a 1-on-1 compatibility with the current protocol first, as the 1.21 code has significantly evolved vs 1.14. I also think today that doing this is the way to go, but am happy to discuss if anyone has a different opinion.