< warren>
meshcollider: meaning "git tag -s" type signatures are safe despite sha1 not being safe?
< warren>
when the news came out I remembered reading somewhere that only git signed commits are safe
< meshcollider>
Oh actually even an annotated tag would just reference the commit by sha1 hash I think, so it would still rely on second preimage resistance
< meshcollider>
Why don't we include the tree-sha512 in the tag just like in the merge commits?
< sipa>
can we?
< sipa>
oh, tags have associated text
< meshcollider>
Yeah they can do, and wumpus usually adds text like "Bitcoin Core 0.18.1 release candidate 1"
< meshcollider>
wumpus: do you usually just tag with git tag -m or is there a script you use
< wumpus>
meshcollider: I simply use git tag -s, but yes I think that's a good idea
< wumpus>
could have a script that automatically makes and signs the tag
< wumpus>
and adds in the treehash
< meshcollider>
It'd be pretty simple to make compared to the GitHub merge script, if you'd be happy to use it
< promag>
jamesob: #16718 also improves -reindex-chainstate right?
< bitcoin-git>
bitcoin/master 403e372 João Barbosa: qa: Relax so that the subscriber is ready before publishing zmq messages
< bitcoin-git>
bitcoin/master 119e97a MarcoFalke: Merge #16740: qa: Relax so that the subscriber is ready before publishing ...
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #16740: qa: Relax so that the subscriber is ready before publishing zmq messages (master...2019-08-slowjoiner) https://github.com/bitcoin/bitcoin/pull/16740
< bitcoin-git>
bitcoin/master e4f4ea4 practicalswift: lint: Catch use of [] or {} as default parameter values in Python functions
< bitcoin-git>
bitcoin/master cc40b55 MarcoFalke: Merge #16726: tests: Avoid common Python default parameter gotcha when mut...
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #16726: tests: Avoid common Python default parameter gotcha when mutable dict/list:s are used as default parameter values (master...python-mutable-default-parameter-values) https://github.com/bitcoin/bitcoin/pull/16726
< gleb>
provoostenator: Do we really want to download asmap.raw needed for testing from somewhere online? Where do you think it belongs, somewhere on github? Maybe in the repo with scripts to build asmap
< provoostenator>
We host a bunch of bigger files on bitcoincore.org
< provoostenator>
Like the macOS SDK
< gleb>
And then, perhaps if file is not available — omit those tests with a warning or something.
< provoostenator>
A torrent could add some redudancy
< gleb>
Sounds like an overkill to me. Alternative source might be the repo with scripts to build asmap.raw, if we end up having a dedicated repo.
< provoostenator>
Yes, if missing the test can be skipped, we have a standard method for that in the functional test suite.
< provoostenator>
Building it manually wouldn't be determinstic, right?
< sipa>
gleb: for testing we don't actually need a big asmap, right?
< sipa>
we could construct a simplified one of a few kilobytes
< provoostenator>
There should be a way for people to verify that the end result that we ship is generated from something that they can at least in principle download / generate.
< gleb>
sipa: Yeah, I had that on mind too.
< MarcoFalke>
There is bitcoin-core/qa-assets
< MarcoFalke>
just saying
< MarcoFalke>
no strong opinion if it is worth the tradeoffs for 1MB of test data
< emilengler>
Is there a way to check which chain is currently in use?
< MarcoFalke>
help getblockchaininfo
< emilengler>
MarcoFalke: I mean a code function which returns the chain name but the rpccall works as well
< gleb>
And also bytes to document that tests adhere to this map :)
< gleb>
This would be enough to show that /16 might now be single-bucket and multi-bucket.
< gleb>
If extra tests are needed, we can expand this, but generally are there any objections to including completely fake mini-asmap for test purposes?
< MarcoFalke>
What would you want to test with the full map anyway?
< gleb>
I don't know, and that's why I'm asking others.
< MarcoFalke>
In fuzzing, failing test cases are often reduced to the smallest failing test case. I guess you can do the same in unit tests for a map that passes the test case :)
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #16745: wallet: Translate all initErrors in CreateWalletFromFile (master...1908-walletTranslate) https://github.com/bitcoin/bitcoin/pull/16745
< promag>
anyone getting "./google/protobuf/stubs/common.h:38:10: fatal error: 'assert.h' file not found" while make in depends?
< fanquake>
promag: Are you building for macOS and missing the SDK?
< promag>
yeah that was the problem
< fanquake>
👍
< promag>
fanquake: have you tried to bump qt@depends?
< fanquake>
promag: No. Will probably take a look at 5.12 in depends post v0.19.0
< fanquake>
There is a PR open with a 5.9.x bump that might go in for 0.19.0. However last time I looked it didn't have any bug fixes that'd benefit us.