< laanwj>
(it's set to moderated, so every post has to be manually accepted, and nothing will come from @bitcoincore.org, just the mail the original sender used)
< bitcoin-git>
bitcoin/master 2ce7f95 fanquake: doc: clean out release notes post branch-off
< bitcoin-git>
bitcoin/master 10eb000 MarcoFalke: Merge bitcoin/bitcoin#22515: doc: clean out release notes post branch-off
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #22515: doc: clean out release notes post branch-off (master...release_note_cleanout_post_22_x) https://github.com/bitcoin/bitcoin/pull/22515
< bitcoin-git>
[bitcoin] darosior opened pull request #22516: macdeploy: alternative info to download the macOS SDK (master...nit_guix_doc) https://github.com/bitcoin/bitcoin/pull/22516
< bitcoin-git>
bitcoin/master cdd51e8 Adrian-Stefan Mares: torcontrol: Resolve Tor control plane address
< bitcoin-git>
bitcoin/master 1c046bb W. J. van der Laan: Merge bitcoin/bitcoin#22288: Resolve Tor control plane address
< bitcoin-git>
[bitcoin] laanwj merged pull request #22288: Resolve Tor control plane address (master...feature/tor-control-dns-resolve) https://github.com/bitcoin/bitcoin/pull/22288
< jonatack>
hm, so after running contrib/guix/guix-build for 20 hours, it blew up at the "common.mkdir /bitcoin/depends/work/build/x86_64-apple-darwin18/boost/1_71_0-9633eb7da14/stage" step, with "./boost/config/detail/select_stdlib_config.hpp:18:12: fatal error: 'cstddef' file not found"
< hebasto>
macos SDK?
< hebasto>
jonatack: ^
< fanquake>
yes that means your missing the macOS sdk
< fanquake>
however Guix is meant to check that it’s available and fail if you’re trying to build sore Darwin and it’s not present
< fanquake>
*for
< jonatack>
thanks. in depends/SDKs i have Xcode-12.1-12A7403-extracted-SDK-with-libcxx-headers/
< hebasto>
looks correct
< jonatack>
ran "contrib/guix/guix-build" from root
< fanquake>
then I’d assume you’re extracted SDK is broken somehow
< jonatack>
ok. i'll retry
< fanquake>
Guix just checks for the existence of the right directory
< hebasto>
to test macos only guix build you could use `env HOSTS=x86_64-apple-darwin18 contrib/guix/guix-build`
< fanquake>
Or just skip Guix and test make -C depends HOST=x86_64-apple-darwin18. It will fail just the same if there's an SDK issue.
< jonatack>
thanks! the first one finds the macOS SDK, so letting it run
< laanwj>
might be useful to compare the directory listing of the SDK to others
< jonatack>
nope, same error
< fanquake>
What do you mean by finds the SDK?
< fanquake>
It will just fail the same as before
< jonatack>
"Found macOS SDK at '/home/jon/projects/bitcoin/bitcoin/depends/SDKs/Xcode-12.1-12A7403-extracted-SDK-with-libcxx-headers', using..."
< fanquake>
It "found" the SDK the first run as well
< laanwj>
it checks that the tar.gz is present but not if it contgains what is expected in it
< laanwj>
oh you didn't extract it from the apple one, but use the pre-extracted one from the CI? i don't understand what could have gone wrong then
< fanquake>
there is so much crap packaged into our SDK tarball that we don't actually need
< fanquake>
like say, 190mb of man pages
< laanwj>
it's magnitudes smaller than the apple download
< fanquake>
yes, certainly still much smaller than xcode, or the command line tools package
< laanwj>
it still contains some deadweight. but dunno if it's worth micro-optimizing
< fanquake>
We could switch to extracting it from the cmd line tools package, save people downloading 10GB, instead they can download ~400mb. Could cull some cruft at that point
< laanwj>
nice
< fanquake>
In any case, I've diffed the CI tarball, to one I have here, and as expected, /usr/include is the same minus metadata
< jonatack>
running env HOSTS=x86_64-apple-darwin18 contrib/guix/guix-build seems to be working this time -- thanks!
< hebasto>
🎉
< jonatack>
(same wget command both times, the extraction may have been bad the first time--i don't know, but good call)
< fanquake>
I mean, it has to have been, otherwise the build would have worked. Would have been interesting to diff the dirs
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #22517: fuzz: Temporarily disable failing assert in banman fuzz test (master...2107-fuzzBanman) https://github.com/bitcoin/bitcoin/pull/22517
< jamesob>
Interesting that we never specifically RaiseValidity to BLOCK_VALID_CHAIN... seems like that only ever happens by way of raising to BLOCK_VALID_SCRIPTS
< bitcoin-git>
[bitcoin] jonatack opened pull request #22520: test: assert on the values of getblockchaininfo time/mediantime (master...getblockchaininfo-time-tests) https://github.com/bitcoin/bitcoin/pull/22520