Awesome, will check.

Please note that many places could be rewritten to use member initializers. And on many places, the diff can be shortened by rewriting initialization process, simplifying this-> stuff and co. But this was not the goal here...

Right, the goal is easy review and checking, as well as being able to enable those warnings by default after this.

Eh, oops I checked the bitcoind binary while this relates to bitcoin-qt, need to recheck.
Ignore the above approval for now (I don't think I can delete it).

My binary comparator finds differences in the following functions:

0000000000000000 <WalletModel::WalletModel(PlatformStyle const*, CWallet*, OptionsModel*, QObject*)>:
0000000000000000 <WalletView::WalletView(PlatformStyle const*, QWidget*)>:
0000000000000000 <BitcoinGUI::BitcoinGUI(PlatformStyle const*, NetworkStyle const*, QWidget*)>:
0000000000000000 <EditAddressDialog::EditAddressDialog(EditAddressDialog::Mode, QWidget*)>:
0000000000000000 <AskPassphraseDialog::AskPassphraseDialog(AskPassphraseDialog::Mode, QWidget*)>:
0000000000000000 <AddressTableModel::AddressTableModel(CWallet*, WalletModel*)>:
0000000000000000 <SignVerifyMessageDialog::SignVerifyMessageDialog(PlatformStyle const*, QWidget*)>:
0000000000000000 <AddressBookPage::AddressBookPage(PlatformStyle const*, AddressBookPage::Mode, AddressBookPage::Tabs, QWidget*)>:
0000000000000000 <PlatformStyle::PlatformStyle(QString const&, bool, bool, bool)>:
0000000000000000 <SendCoinsEntry::SendCoinsEntry(PlatformStyle const*, QWidget*)>:
0000000000000000 <RPCConsole::RPCConsole(PlatformStyle const*, QWidget*)>:

These are just the constructors, even empty ones. I don't see why, it may have something to do with Qt's moc generating different metadata?

ut/code review ACK 46b15bb

Edit: new commit removes two functions from the list

Would be interesting to know why the binaries don't match for qt. Encountered the same here: #8658 (comment)

The difference cause by this PR is probably different from the difference caused by #8658.

In this PR, the change in 46b15bb illustrates the reason. Technically it does the same, but compiler has two places where to get the value used (member variable initialized from the argument or the argument itself).

@MarcoFalke yes we figured it out on IRC: it has to do with name binding, using the argument versus member variable it was assigned to. For example this seems to make the differences in AddressBookPage::AddressBookPage go away: http://www.hastebin.com/jixoyufuxu.patch .

But that's kind of ugly. So in this case I prefer slightly different executables to larger source code patch that makes the code uglier. If it was consensus code it'd be different.

Squashed the squashme commit.