Wed, Jun 9
Tue, Jun 8
you're right, this is not good
Note to reviewers:
This supersedes D9489 and D9493. I deliberately chose to not split apart the filtering of the peer's known proof (was D9493) because in the end I don't think it makes the review easier, but quite the opposite. By keeping it together with the inv message sending it makes it easier to compare with what is done for the proofs invs.
The broadcasting feature detailed in D9489 will be added in another diff.
Rebase on top of D9648, make it an empty shell for now
Unrelated failure, I am out of luck today
Rebase on D9649 to get more factorized code
Confused with another diff
Unrelated spurious processor_tests failure. I managed to reproduce by chance on my machine and I have a core dump to debug on now
Mon, Jun 7
Use the right clock for the job
Why not have two flags that can be set, -uaclientname, and -uaclientversion ?
Changed FormatSubVersion to FormatUserAgent and solely use FormatSubVersionUserAgent now
I don't know what caused the failure, I couldn't reproduce in a 100 loop on a similar debian, nor did it reproduce on CI.
Added the rational in the diff summary
Investigating the build failure
rebase (unrelated failure)
rebase to fix clang failure
fix clang warning
Can you expand a bit of what the problem is from setting the flag globally ? Without more details it's hard to see how having the option repeated in and here in every test is better than setting it once for all.
Also if it increases the test coverage then the tests are wrong, because these tests are not testing the -ecash flag.
Sun, Jun 6
Changed to -uaclient, which sets the client name and version of the user agent, staying compatible with -uacomment
couldn't reproduce this failure. rebasing on latest master.
bummer. alright, how about D9638, then?
rebase on master