Philip Kaufmann [Mon, 5 Nov 2012 07:04:21 +0000 (08:04 +0100)]
update CClientUIInterface and remove orphan Wx stuff
- fix ThreadSafeMessageBox always displays error icon
- allow to specify MSG_ERROR / MSG_WARNING or MSG_INFORMATION without a
custom caption / title
- allow to specify CClientUIInterface::ICON_ERROR / ICON_WARNING and
ICON_INFORMATION (which is default) as message box icon
- remove CClientUIInterface::OK from ThreadSafeMessageBox-calls, as
the OK button will be set as default, if none is specified
- prepend "Bitcoin - " to used captions
- rename BitcoinGUI::error() -> BitcoinGUI::message() and add function
documentation
- change all style parameters and enum flags to unsigned
- update code to use that new API
- update Client- and WalletModel to use new BitcoinGUI::message() and
rename the classes error() method into message()
- include the possibility to supply the wanted icon for messages from
Client- and WalletModel via "style" parameter
Pieter Wuille [Sat, 24 Nov 2012 13:26:51 +0000 (14:26 +0100)]
Bugfix: remove conflicting transactions from memory pool
When a transaction A is in the memory pool, while a transaction B
(which shares an input with A) gets accepted into a block, A was
kept forever in the memory pool.
This commit adds a CTxMemPool::removeConflicts method, which
removes transactions that conflict with a given transaction, and
all their children.
This results in less transactions in the memory pool, and faster
construction of new blocks.
Philip Kaufmann [Sun, 18 Nov 2012 10:58:32 +0000 (11:58 +0100)]
Bitcoin-Qt: fix crash on Windows caused by CDBEnv::EnvShutdown()
- can be triggerd by just adding -proxy=crashme with 0.7.1
- crash occured, when AppInit2() was left with return false; after the
first call to bitdb.open() (Step 6 in init)
- this is caused by GetDataDir() or .string() in CDBEnv::EnvShutdown()
called via the bitdb global destructor
- init fDbEnvInit and fMockDb to false in CDBEnv::CDBEnv()
Pieter Wuille [Tue, 13 Nov 2012 22:03:25 +0000 (23:03 +0100)]
Introduce script verification flags
These flags select features to be enabled/disabled during script
evaluation/checking, instead of several booleans passed along.
Currently these flags are defined:
* SCRIPT_VERIFY_P2SH: enable BIP16-style subscript evaluation
* SCRIPT_VERIFY_STRICTENC: enforce strict adherence to pubkey/sig encoding standards.
o Remove unused Leave and GetLock functions
o Make Enter and TryEnter private.
o Simplify Enter and TryEnter.
boost::unique_lock doesn't really know whether the
mutex it wraps is locked or not when the defer_lock
option is used.
The boost::recursive_mutex does not expose this
information, so unique_lock only infers this
knowledge. When taking the lock is defered, it
(randomly) assumes that the lock is not taken.
boost::unique_lock has the following definition:
Philip Kaufmann [Sun, 11 Nov 2012 12:11:42 +0000 (13:11 +0100)]
some small fixes for main.cpp/.h
- remove an unwanted ";" at the end of the ~CCoinsView() destructor
- in FindBlockPos() and FindUndoPos() only call fclose(), is file is open
- fix an error string in the CBlockUndo class
Philip Kaufmann [Fri, 9 Nov 2012 21:51:40 +0000 (22:51 +0100)]
simplify CDBEnv::Open() / fix small glitches
- remove pathEnv from CDBEnv, as this attribute is not needed
- change path parameter in ::Open() to a reference
- make nDbCache variable an unsigned integer
- remove a missplaced ";" behin ::IsMock()
Philip Kaufmann [Mon, 3 Sep 2012 13:54:47 +0000 (15:54 +0200)]
allow listening on -bind=address for blocked networks
- this allows the client to listen on via -bind specified addresses
(e.g. 127.0.0.1), even when a network (IPv4 in that case) was blocked
via e.g -onlynet="Tor"
- introduce enum BindFlags to avoid passing multiple bools to Bind()
- make -bind help text clear we ALWAYS listen on the specified address
- remove an unused variable
- remove 2 unneeded IsLimited() checks before calling Bind(), which does
these checks anyway
- usage case: specify -bind=127.0.0.1 -onlynet="Tor" to allow incoming
connections to a Tor hidden service, but still don't allow other IPv4
nodes to connect / get connected
Philip Kaufmann [Thu, 8 Nov 2012 18:38:49 +0000 (19:38 +0100)]
don't use memset() in privacy/security relevant code parts
As memset() can be optimized out by a compiler it should not be used in
privacy/security relevant code parts. OpenSSL provides the safe
OPENSSL_cleanse() function in crypto.h, which perfectly does the job of
clean and overwrite data.
For details see: http://www.viva64.com/en/b/0178/
- change memset() to OPENSSL_cleanse() where appropriate
- change a hard-coded number from netbase.cpp into a sizeof()
Philip Kaufmann [Thu, 8 Nov 2012 20:45:32 +0000 (21:45 +0100)]
Qt: small header changes / fixes
- ensure header inclusion guard is named after the header file
- add missing comments at the end of some inclusion guards
- add a small Qt5 compatibility fix in macdockiconhandler.h
grimd34th [Mon, 5 Nov 2012 02:00:57 +0000 (21:00 -0500)]
VariantClear@4 + shutdown@8 patch
merged 2 windows:LIBS lines so they happen when they need to (#976)
added -loleaut32 to fix VariantClear
moved -lws2_32 and others to existing windows:LIBS line
Jeff Garzik [Sun, 4 Nov 2012 22:16:46 +0000 (17:16 -0500)]
RPC: HTTP server uses its own ReadHTTPRequestLine()
rather than reusing ReadHTTPStatus() from the client mode.
The following additional HTTP request validations are added, both in line with
existing HTTP client practice:
1) HTTP method must be GET or POST. Most clients use POST, some
use GET. Either way, this continues to work.
2) HTTP URI must start with "/" character.
Normal URI is "/" (a 1-char string), so this is fine.
Jeff Garzik [Sun, 4 Nov 2012 21:06:38 +0000 (16:06 -0500)]
RPC, cosmetic: push down ReadHTTPStatus() calls into ReadHTTP() callers
ReadHTTPStatus() is currently overloaded: In client mode, it properly parses
and receives an HTTP status line. In server mode, it incorrectly parses the
HTTP request line as an HTTP status line.
This server mode bug has never mattered, because the RPC server never
cared about the URI (path) provided in the HTTP request. That will change in
the future, so go ahead and begin fixing the problem.
This patch is cosmetic, and should result in NO behavior changes.
Further renames:
ReadHTTPHeader -> ReadHTTPHeaders
ReadHTTP -> ReadHTTPMessage
Pieter Wuille [Thu, 1 Nov 2012 14:52:25 +0000 (15:52 +0100)]
Bugfix: do not keep relaying spent wallet transactions
The original test (checking whether the transaction occurs in the
txindex) is not usable anymore, as it will miss anything already
fully spent. However, as merkle transactions (and by extension,
wallet transactions) track which block they were last seen being
included in, we can use that to determine the need for
rebroadcasting.
Philip Kaufmann [Fri, 12 Oct 2012 14:08:47 +0000 (16:08 +0200)]
Bitcoin-Qt: use statustips in addition to tooltips
- add setStatusTip() in addition to setTooltip() where it makes sense
- add only setStatusTip() if GUI element is only used in main- or tray menu
- add an event filter on our BitcoinGUI object to prevent garbelled text
on the status bar, which happens when we use it for e.g. displaying
block-sync state and then a QEvent::StatusTip wants to write own text to it