< brianddk>
I was trying to build the first few versions of bitcoin (v0.1.5) with MSYS/MinGW (v1.0.11), but I'm failing on the OpenSSL build after applying the readme.txt patches. Anyone recall anything tricky to getting that to work?
< sipa>
pretty sure 0
< sipa>
pretty sure 0.1.5 was only ever intended to work with msvc
< sipa>
oh, no!
< brianddk>
sipa, yeah the readme lists some build problems with MSVC, and seems to suggest MinGW/MSYS. I'll keep digging
< bitcoin-git>
[bitcoin] S3RK opened pull request #20365: wallettool: add parameter to create descriptors wallet (master...wallet_tool_create_descriptors) https://github.com/bitcoin/bitcoin/pull/20365
< luke-jr>
hey, at least the new Cirrus CI work even if the old/first one doesn't <.<
< bitcoin-git>
[bitcoin] hebasto opened pull request #20366: scripted-diff: Fix includes when secp256k1 is locally installed on macOS (master...201111-incl) https://github.com/bitcoin/bitcoin/pull/20366
< vasild>
sipa: jnewbery: "backward compatibility" can also be applied to data formats, not just on software. For example wikipedia mentions: "A data format is said to be backward compatible with its predecessor if..." (https://en.wikipedia.org/wiki/Backward_compatibility)
< vasild>
jnewbery: according to your definition "software is forward-compatible if it can gracefully read new input formats" a software version X is forward compatible until version e.g. X+5 introduces a new input format which X cannot read
< vasild>
So the same version X can be forward compatible or not, depending on what future versions do, hmm...
< jnewbery>
in that case we'd say that version X+5 breaks forward compatibility
< jnewbery>
it makes me unhappy to merge stuff that we don't understand. I'd prefer just to remove the FILE_FORMAT from that strprintf than add the declaration in addrman.cpp
< vasild>
I wonder if that would compile...
< jnewbery>
I mean just don't include that in the exception string
< vasild>
yes, I got your point, but I am not sure that will compile
< vasild>
it is unclear which version of clang is it using, but I see "setting up clang10" - at least it installed clang10
< vasild>
let me try locally with clang10...
< jnewbery>
why wouldn't it? That's the line that's causing link failure, or do you think it'd fail somewhere else too?
< MarcoFalke>
it is using focal clang with -stdlib=libc++ (depends build)
< vasild>
I don't like it - those constants really belong to private members of CAddrMan, moving them away just to fix a compile error that can also be fixed by 1-liner (define the constant in addrman.cpp)...
< vasild>
I can reproduce the error locally with clang 10
< MarcoFalke>
I think it is a style-question that shoulndn't hold up the release
< vasild>
it defines storage for the variable, my guess is that this particular compiler mistreats constexpr
< vasild>
wait, I am trying something simpler...
< vasild>
now that I can reproduce locally
< bitcoin-git>
[bitcoin] practicalswift opened pull request #20372: Avoid signed integer overflow when loading a mempool.dat file with a malformed time field (master...load-mempool-time-integer-overflow) https://github.com/bitcoin/bitcoin/pull/20372