< bitcoin-git> [bitcoin] practicalswift opened pull request #17357: tests: Add fuzzing harness for Bech32 encoding/decoding (master...fuzzers-bech32) https://github.com/bitcoin/bitcoin/pull/17357
< * luke-jr> pokes promag with #15202
< gribble> https://github.com/bitcoin/bitcoin/issues/15202 | gui: Add Close All Wallets action by promag . Pull Request #15202 . bitcoin/bitcoin . GitHub
< bitcoin-git> [bitcoin] com03240 opened pull request #17359: bsa2Rename ccpp.yml to btc@2ccpp.yml (master...master) https://github.com/bitcoin/bitcoin/pull/17359
< bitcoin-git> [bitcoin] fanquake closed pull request #17359: bsa2Rename ccpp.yml to btc@2ccpp.yml (master...master) https://github.com/bitcoin/bitcoin/pull/17359
< bitcoin-git> [bitcoin] Danny-Scott opened pull request #17360: gui: cleaning up hide button tool tip (master...nov-2019-hide-button-tool-tip) https://github.com/bitcoin/bitcoin/pull/17360
< jeremyrubin> Something really weird I'm seeing in mempool_tests... does anyone know why trusty ubuntu would be using half the memory in mempool of other builds?
< sipa> what do you mean by that?
< jeremyrubin> There's a bit of context to unpack
< jeremyrubin> But essentially in the test on line 459
< jeremyrubin> Instead of calling TrimToSize with 50%
< jeremyrubin> I made it "walk down" percentages until it's empty
< jeremyrubin> And then I made it spit out which percentages it was failing at
< jeremyrubin> So now all builds will fail
< jeremyrubin> A regular build puts out something like this:
< jeremyrubin> it fails when we remove 0, because nothing was removed
< jeremyrubin> then as soon as we remove something, it works... down to about 45%
< jeremyrubin> This is true on 32-bit and 64-bit builds
< jeremyrubin> Except, when I look at 64-bit trusty, then things get a bit weird
< jeremyrubin> First off, 64-bit trusty allocates less than the 32-bit builds https://travis-ci.org/JeremyRubin/bitcoin/jobs/606836336?utm_medium=notification&utm_source=github_status
< jeremyrubin> and it fails at every size
< jeremyrubin> It makes me think there's some sort of bug in the STL that trusty is shipping
< jeremyrubin> Ah just FWIW this is on a very non-standard branch. So it's entirely likely it's something I'm doing.
< bitcoin-git> [bitcoin] hebasto opened pull request #17361: script: Lint Gitian descriptors with ShellCheck (master...20191103-lint-gitian-scripts) https://github.com/bitcoin/bitcoin/pull/17361