(1) If coverage is outlined as “every part that is not consensus”, but these coverage checks are included within the consensus recordsdata… what precisely are they doing there? Are they deprecated?
No, not deprecated. There is not a clear dividing line within the Bitcoin Core file system the place say one file solely checks coverage guidelines and one file solely checks consensus guidelines. (See this StackExchange submit from Suhas Daftuar discussing consensus and coverage subdirectories). That is for numerous causes. One is the inheritance of the codebase from Satoshi who created a big bundle of code that though refactored and cleaned up over a decade by Core contributors nonetheless bears a resemblance to what they inherited from Satoshi. Maybe extra importantly it’s actually tough to stipulate what all of the consensus guidelines really are as a result of there are edge circumstances the place you assume the code you are touching is not consensus associated however then it seems you launched a consensus bug that ends in a possible chain break up. Pieter Wuille mentioned this on the Chaincode Labs podcast in 2020:
One of many issues I believe realized from that’s specifying what your consensus guidelines is basically laborious. That doesn’t imply you possibly can’t strive however who would’ve thought {that a} configuration setting within the database layer you might be utilizing really leaked semantically into Bitcoin’s implicitly outlined consensus guidelines. You possibly can attribute that to human failure in fact. We must always’ve learn the documentation and been conscious of that.
There could possibly be a bug in your C library. There could be a bug in your kernel. There may even be a bug in your CPU.
We will speak in regards to the boundary in making an attempt to summary the a part of the codebase that deliberately contributes to consensus however it is vitally laborious to say clearly this code has no affect on consensus code as a result of bugs can leak. I believe one of many issues to be taught there’s you really need software program that’s meant to be used in a consensus system the place not solely you will have the requirement that if everybody behaves appropriately all people accepts the suitable reply but additionally that everyone will disagree about what’s an invalid piece of information in lockstep.
(2) Are ALL of the consensus guidelines included someplace in these consensus recordsdata? I am assuming not. So, I ask…
No. There have been makes an attempt prior to now to create a consensus library (libbitcoinconsensus) and extra just lately libbitcoinkernel that Carl Dong has mentioned and contributed to however it’s actually laborious. Consensus is a very slippery idea in Bitcoin.
(3) May we compile an inventory, maybe, of every part {that a} legitimate block would wish to be able to be accepted as legitimate that’s NOT already included someplace in these consensus recordsdata??
When you imply like an exhaustive listing within the English language then you are going to wrestle. You possibly can strive and you’ll in all probability embody many of the apparent issues however it’s the edge circumstances and issues that appear they’re solely non consensus associated that end up to not be which might be lacking.
