< fanquake>
wumpus only 2 signed sigs? Or haven't you pushed yours up
< wumpus>
yes, I still have to upload mine
< wumpus>
thanks for remindingm e
< fanquake>
np
< provoostenator>
I'm rebuilding my unsigned ones first because it was complaining about "cp: cannot stat 'inputs/bitcoin-0.17.1rc1-win-unsigned.tar.gz"
< provoostenator>
(just copied a neweer version of gitian-build.py
< provoostenator>
Happy to review 13646 again. Can someone remind me of the git incantation to see what changed relative to my 94677cc utACK? (assuming I still have it locally)
< wumpus>
if you just want to see the difference since then: git diff 94677cc
< wumpus>
this will work okay if either commits were added since then, or commits were squashed/reorganized in place. As it is the total diff that includes the difference in base commits it's not useful to review a rebase.
< wumpus>
for reviewing a rebase*
< fanquake>
Chris_Stewart_5 is rapidcheck work essentially blocked on upstream?
< Chris_Stewart_5>
fanquake: I haven't had time to track down the root cause. Does it appear that memory access violation is happening in rapidcheck?
< fanquake>
Chris_Stewart_5 Yes I think so. The bitcoin related test which is failing seems too simple? to be the cause.
< fanquake>
Chris_Stewart_5 possibly, if you just want to be able to merge #14853 (assuming no other tests fail) in a similar way. I can test later tonight.
< gribble>
https://github.com/bitcoin/bitcoin/issues/14853 | [wip] depends: latest rapidcheck, enable property based tests on Travis by fanquake · Pull Request #14853 · bitcoin/bitcoin · GitHub
< fanquake>
When I tried compiling #14430 I ran into some other, potentially related, issues.
< gribble>
https://github.com/bitcoin/bitcoin/issues/14430 | Add more property based tests for basic bitcoin data structures by Christewart · Pull Request #14430 · bitcoin/bitcoin · GitHub
< Chris_Stewart_5>
fanquake: I believe the issues with #14430 where from not having the EXTRAS module installed from rapidcheck? At least I believe that was MarcoFalke's problem with #14171
< gribble>
https://github.com/bitcoin/bitcoin/issues/14430 | Add more property based tests for basic bitcoin data structures by Christewart · Pull Request #14430 · bitcoin/bitcoin · GitHub
< Chris_Stewart_5>
fanquake: Actually looking closer it appears there are two different properties failing between 14171 and 14430 'key_sign_symmetry' and 'key_set_symmetry'
< Chris_Stewart_5>
Interesting, i missed that comment
< Chris_Stewart_5>
did we bump to c++14?
< wumpus>
Chris_Stewart_5: no
< fanquake>
Chris_Stewart_5 No, maybe for 0.19.0 #13356
< fanquake>
A lot of my vm setup/testing had been pretty manual using virtualbox, so I've decided to try and automate it further.
< fanquake>
For ex, make it's fairly easy to github-merge a PR, then "spin up" an openBSD vm, with core dependencies/debugging tools already available, and have the code sitting inside.
< Chris_Stewart_5>
That can be used for more generalized testing, i.e. rapid check stuff right? Is there much of a difference between this and using docker?
< provoostenator>
wumpus: thanks. Indeed I want to review the rebase, which involves slightly more magic
< fanquake>
Chris_Stewart_5 Yea. I did use Docker directly to recreate Travis for the rapidcheck debugging, but decided to use vagrant here, as it sits "above" virtualbox, docker etc
< Chris_Stewart_5>
Interesting. I probably won't get around to testing until this weekend but I'll give it a try.
< provoostenator>
Stack Overflow to the rescue: git commit-tree HEAD^{tree} -p 94677cc -p HEAD~5 -m "Rebase changes since 94677cc" && git show
< provoostenator>
Where 5 is the number of commits in the PR
< provoostenator>
(neh, I don't think that worked, I'll leave a comment on the PR once I find it)
< provoostenator>
Ah, it's: git show ` git commit-tree ...`
< fanquake>
If you do a github-merge 14853, then vagrant up travis-linux-no-depends-qt
< fanquake>
At the end you can ssh into the box, and run gdb --args src/test/test_bitcoin --log_level=all --run_test=key_properties
< fanquake>
then "run", "thread apply bt all" and it'll give you the backtrace you're after
< wumpus>
fanquake: that seems very useful, I'm still mostly using manual testing in VMs too, can be a hassle sometimes especially when the VMs need to be upgraded, would be good to automate
< wumpus>
provoostenator: good to know
< provoostenator>
wumpus: actually that didn't work either. I left a git range-diff incantation on the PR, though the diff is too big to be useful.
< wumpus>
provoostenator: what I usually do is manually repeat the rebase (on a temporary branch), then compare against the author's, it's a lot of hassle though especially the rebase doesn't go cleanup
< wumpus>
cleanly*
< wumpus>
I'd hoped there was a cleaner/faster way but seems not
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #14925: test: Correctly deserialize without witness (master...Mf1812-testWitnessDeser) https://github.com/bitcoin/bitcoin/pull/14925
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #14926: test: consensus: Check that final transactions are valid (master...Mf1812-testConsensusFinal) https://github.com/bitcoin/bitcoin/pull/14926