Fix gui segfault caused by bitcoin/bitcoin#22216 #361
Conversation
Reported by Hennadii Stepanov bitcoin/bitcoin#22216 (comment) Fixes bitcoin/bitcoin#22227
|
Why tests and CI did not catch it? |
|
Tests would not catch it because tests initialize context differently than the gui app. I didn't catch it because I only tested the gui with bitcoin/bitcoin#22216 and bitcoin/bitcoin#22219 together. They were initially part of the same PR and bitcoin/bitcoin#22219 also fixes this bug a different way. |
There was a problem hiding this comment.
ACK d7f3b1a
Tested on macOS 11.3, Qt 5.15.2
master:
xyz@xyzs-MBP bitcoin % ./src/qt/bitcoin-qt -testnet
dbus[52399]: Dynamic session lookup supported but failed: launchd did not provide a socket path, verify that org.freedesktop.dbus-session.plist is loaded!
Assertion failed: ("node.args" && check), function operator(), file wallet/init.cpp, line 127.
zsh: abort ./src/qt/bitcoin-qt -testnet
pr:
Runs and works fine upon manual testing
d7f3b1a Fix gui segfault caused by bitcoin#22216 (Russell Yanofsky) Pull request description: Reported by Hennadii Stepanov bitcoin#22216 (comment) Fixes bitcoin#22227 ACKs for top commit: hebasto: ACK d7f3b1a, tested on Linux Mint 20.1 (Qt 5.12.8). jarolrod: ACK d7f3b1a Tree-SHA512: d672bfa9f1bcd500a879ec7ed27096086ae93b73ad5da8090f29cc5b6d985c46a76583cc384304d67210f87b6b839c2391f0fcc24fd3588c4a014e540283fdfe
Is there a way to improve tests so that they initialize in the same way? |
One way would be to make the python tests invoke bitcoin-qt. I believe this kind of already works if you set |
Reported by Hennadii Stepanov bitcoin/bitcoin#22216 (comment)
Fixes bitcoin/bitcoin#22227