When submitpackage produced no per-transaction result for a member,
the RPC previously set "error": "unevaluated" but then continued
without inserting the entry into tx-results, making it impossible for
callers to know which wtxids were unevaluated.
Insert the placeholder result before continuing, update help text, and
adjust functional tests to expect entries for all submitted wtxids.
Throw RPC_CLIENT_INVALID_IP_OR_SUBNET when LookupHost(addr, false) fails
in addpeeraddress. This aligns with setban/addconnection and avoids the
opaque {"success": false} result for input errors. The JSON {success,
error?} object remains for addrman outcomes only. Update test to match.
This checks whether the ARMv8.5 extensions RNDR and RNDRRS
are available and uses them for random entropy purposes.
They are functionally identical to the x86 RDRAND/RDSEED
extensions and are used in a similar manner.
Starting bitcoin-qt with non-dash ("-") arguments causes it to
silently ignore any later valid options. This change makes the
client exit with an error message if any such "loose" arguments
are encountered.
However, allow BIP-21 'bitcoin:' URIs only if no other options
follow.
If a user opens multiple wallets in the GUI from the
menu bar, the last one opened is the active one in
the main window. However, For the RPC Console window,
the _first_ one opened is active. This can be
confusing, as wallet RPC commands may be sent to a
wallet the user didn't intend.
This commit makes the RPC Console switch to the wallet
opened from the menu bar.
GetErrorReason()'s Win32 implementation does the same thing as
Win32ErrorString(int err) from syserror.cpp, so call the latter.
Also remove now-unnecessary headers from sock.cpp and less verbose
handling of #ifdefs.
Only raw errno codes are logged if FileCommit fails. These are
implementation-specific, so it makes it harder to debug based on
user reports. Instead, use SysErrorString to display both the
raw int value and the descriptive message.
The current `FormatISO8601DateTime` function will
return an empty string if it encounters an error
when converting the `int64_t` seconds since epoch
to a formatted date time. In the unlikely case that happens,
`strStamped.pop_back()` would be undefined behavior.
Since users may have thought the null characters in their
passphrases were actually evaluated prior to this change,
they may be surprised to learn that their passphrases no
longer work. Give them feedback to explain how to remedy
the issue.
`SecureString` is a `std::string` specialization with
a secure allocator. However, it's treated like a C-
string (no explicit length and null-terminated). This
can cause unexpected behavior. For instance, if a user
enters a passphrase with an embedded null character
(which is possible through Qt and the JSON-RPC), it will
ignore any characters after the null, giving the user
a false sense of security.
Instead of assigning `SecureString` via `std::string::c_str()`,
assign it via a `std::string_view` of the original. This
explicitly captures the size and doesn't make any extraneous
copies in memory.
When an encrypted wallet is locked (for instance via the
RPC `walletlock`), the docs indicate that the key is
removed from memory. However, the vector (with a secure
allocator) is merely cleared. This allows the key to persist
indefinitely in memory. Instead, manually fill the bytes with
zeroes before clearing.
One test case uses snprintf to convert an
int to a string. Change it to use ToString
(which uses a locale-independent version of
std::to_string). Also remove unnecessary
parts of StringContentsSerializer.
The way that the main overview page limits the number
of transactions displayed (currently 5) is not
an appropriate use of Qt. If it's run with a DEBUG
build of Qt, it'll result in a segfault in certain
relatively common situations. Instead of artificially
limiting the rowCount() in the subclassed proxy
filter, we hide/unhide the rows in the displaying
QListView upon any changes in the sorted proxy filter.
In certain circumstances, the GUI console will display
the message 'Executing command without any wallet' when
it is, in fact, using the default wallet.
In RPC calls, if no wallet is explicitly selected and
there is exactly one wallet loaded, the default is to
act on that loaded wallet.
The GUI console acts that way in reality, but
erroneously reports that it's not acting on any
particular wallet.
If you open a wallet and send a shutdown signal during
that process, the GUI will segfault due to some queued
wallet events happening after the wallet controller
is deleted. This is a minimal fix for those issues.
To be eligible for fee-bumping, a transaction must not have any
of its outputs (eg - change) spent in other unconfirmed transactions
in the wallet. However, this check should not apply to abandoned
transactions.
A new test case is added to cover this case.
This fully closes bitcoin#12179. Currently, in the GUI, when a user
abandons a transaction, a call is made to remove it from the list,
and another signal fires (eventually) that adds it back to the GUI
with a trash can icon.
There are no conditions where the abandoned transaction should be
directly removed from the GUI. If the underlying model changes, the
deletion will be reflected anyway.
MacOS 13 sends a window focus change notification after the main
window has been destroyed but before the QTApplication has been
destroyed. This results in the menu bar receiving a notification
despite it no longer existing. The solution is to pass the main
window as context when subscribing to the notifications. Qt
automatically unsubscribes to notifications if the sender OR
context is destroyed.
John Moffett