Details
- Reviewers
deadalnix jasonbcox - Group Reviewers
Restricted Project - Commits
- rSTAGING661d4af11a08: Declare single-argument (non-converting) constructors "explicit"
rABC661d4af11a08: Declare single-argument (non-converting) constructors "explicit"
make check
Diff Detail
- Repository
- rABC Bitcoin ABC
- Branch
- arcpatch-D2207
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 4338 Build 6741: Bitcoin ABC Buildbot (legacy) Build 6740: arc lint + arc unit
Event Timeline
reverse_iterator.h and qt/callback.h have changes in the backport, but these files do not exist in ABC. Are there backports we need that this depends on?
Also missing explicit on ConnectTrace constructor in validation.cpp
src/blockencodings.h | ||
---|---|---|
226 ↗ | (On Diff #6358) | Missed one |
src/chain.h | ||
250 ↗ | (On Diff #6358) | Remove this. Looks accidentally added. |
src/netaddress.h | ||
39 ↗ | (On Diff #6358) | Missed |
103 ↗ | (On Diff #6358) | Missed |
src/qt/utilitydialog.h | ||
42 ↗ | (On Diff #6358) | nullptr change isn't part of the backport, but looks fine to me |
src/support/lockedpool.h | ||
238 ↗ | (On Diff #6358) | Missed |
src/tinyformat.h | ||
166 ↗ | (On Diff #6358) | Is there a missing backport for format_error? |
src/wallet/wallet.h | ||
720 ↗ | (On Diff #6358) | This should be explicit. The backport has a similar, but not identical, constructor |
tinyformat format_error:
The missing backport is PR9963, which addresses the consequences of a format error. IMO this is not a good idea because it does not solve the root issue, the bad format itself (if you remember, some have been fixed recently in addrman). A later PR propose a linter to avoid the format issues which seems by far a better option to me.
qt/callback.h
The missing backport is PR10098, which fixes QT4 compatibility issues. This is reversed in PR14527 when support for QT4 is dropped, so there is no reason backport PR10098.
reverse_iterator.h
The missing backport is PR10193, which is intended to remove BOOST_FOREACH. The method actually used in the codebase is boost::adaptors::reverse. I don't really see any advantage in using a custom reverse_iterator implementation here, as boost is already used everywhere in the code.
src/blockencodings.h | ||
---|---|---|
226 ↗ | (On Diff #6358) | This backport is only about single argument constructors |
src/qt/utilitydialog.h | ||
42 ↗ | (On Diff #6358) | Yes it's part of D2206, which is supposed to be a dependency... I may have missed a rebase |