* Fix AMM offer rounding and low quality LOB offer blocking AMM:
A single-path AMM offer with account offer on DEX, is always generated
starting with the takerPays first, which is rounded up, and then
the takerGets, which is rounded down. This rounding ensures that the pool's
product invariant is maintained. However, when one of the offer's side
is XRP, this rounding can result in the AMM offer having a lower
quality, potentially causing offer generation to fail if the quality
is lower than the account's offer quality.
To address this issue, the proposed fix adjusts the offer generation process
to start with the XRP side first and always rounds it down. This results
in a smaller offer size, improving the offer's quality. Regardless if the offer
has XRP or not, the rounding is done so that the offer size is minimized.
This change still ensures the product invariant, as the other generated
side is the exact result of the swap-in or swap-out equations.
If a liquidity can be provided by both AMM and LOB offer on offer crossing
then AMM offer is generated so that it matches LOB offer quality. If LOB
offer quality is less than limit quality then generated AMM offer quality
is also less than limit quality and the offer doesn't cross. To address
this issue, if LOB quality is better than limit quality then use LOB
quality to generate AMM offer. Otherwise, don't use the quality to generate
AMM offer. In this case, limitOut() function in StrandFlow limits
the out amount to match strand's quality to limit quality and consume
maximum AMM liquidity.
The AMM has an invariant for swaps where:
new_balance_1*new_balance_2 >= old_balance_1*old_balance_2
Due to rounding, this invariant could sometimes be violated (although by
very small amounts).
This patch introduces an amendment `fixAMMRounding` that changes the
rounding to always favor the AMM. Doing this should maintain the
invariant.
Co-authored-by: Bronek Kozicki
Co-authored-by: thejohnfreeman
* It is now an invariant that all constructed Public Keys are valid,
non-empty and contain 33 bytes of data.
* Additionally, the memory footprint of the PublicKey class is reduced.
The size_ data member is declared as static.
* Distinguish and identify the PublisherList retrieved from the local
config file, versus the ones obtained from other validators.
* Fixes#2942
This reverts commit 002893f280.
There were two files with conflicts in the automated revert:
- src/ripple/rpc/impl/RPCHelpers.h and
- src/test/rpc/JSONRPC_test.cpp
Those files were manually resolved.
* Remove include <ranges>
* Formatting fix
* Output for subscriptions
* Output from sign, submit etc.
* Output from ledger
* Output from account_tx
* Output from transaction_entry
* Output from tx
* Store close_time_iso in API v2 output
* Add small APIv2 unit test for subscribe
* Add unit test for transaction_entry
* Add unit test for tx
* Remove inLedger from API version 2
* Set ledger_hash and ledger_index
* Move isValidated from RPCHelpers to LedgerMaster
* Store closeTime in LedgerFill
* Time formatting fix
* additional tests for Subscribe unit tests
* Improved comments
* Rename mInLedger to mLedgerIndex
* Minor fixes
* Set ledger_hash on closed ledger, even if not validated
* Update API-CHANGELOG.md
* Add ledger_hash, ledger_index to transaction_entry
* Fix validated and close_time_iso in account_tx
* Fix typos
* Improve getJson for Transaction and STTx
* Minor improvements
* Replace class enum JsonOptions with struct
We may consider turning this into a general-purpose template and using it elsewhere
* simplify the extraction of transactionID from Transaction object
* Remove obsolete comments
* Unconditionally set validated in account_tx output
* Minor improvements
* Minor fixes
---------
Co-authored-by: Chenna Keshava <ckeshavabs@gmail.com>
Add new transaction submission API field, "sync", which
determines behavior of the server while submitting transactions:
1) sync (default): Process transactions in a batch immediately,
and return only once the transaction has been processed.
2) async: Put transaction into the batch for the next processing
interval and return immediately.
3) wait: Put transaction into the batch for the next processing
interval and return only after it is processed.
* In namespace ripple, introduces get_name function that takes a
std:🧵:native_handle_type and returns a std::string.
* In namespace ripple, introduces get_name function that takes a
std::thread or std::jthread and returns a std::string.
* In namespace ripple::this_thread, introduces get_name function
that takes no parameters and returns the name of the current
thread as a std::string.
* In namespace ripple::this_thread, introduces set_name function
that takes a std::string_view and sets the name of the current
thread.
* Intended to replace the beast utilities setCurrentThreadName
and getCurrentThreadName.
Add new transaction submission API field, "sync", which
determines behavior of the server while submitting transactions:
1) sync (default): Process transactions in a batch immediately,
and return only once the transaction has been processed.
2) async: Put transaction into the batch for the next processing
interval and return immediately.
3) wait: Put transaction into the batch for the next processing
interval and return only after it is processed.
Use the most recent versions in ConanCenter.
* Due to a bug in Clang 16, you may get a compile error:
"call to 'async_teardown' is ambiguous"
* A compiler flag workaround is documented in `BUILD.md`.
* At this time, building this with gcc 13 may require editing some files
in `.conan/data`
* A patch to support gcc13 may be added in a later PR.
---------
Co-authored-by: Scott Schurr <scott@ripple.com>
- Previously, mulDiv had `std::pair<bool, uint64_t>` as the output type.
- This is an error-prone interface as it is easy to ignore when
overflow occurs.
- Using a return type of `std::optional` should decrease the likelihood
of ignoring overflow.
- It also allows for the use of optional::value_or() as a way to
explicitly recover from overflow.
- Include limits.h header file preprocessing directive in order to
satisfy gcc's numeric_limits incomplete_type requirement.
Fix#3495
---------
Co-authored-by: John Freeman <jfreeman08@gmail.com>
Misaligned load and store operations are supported by both Intel and ARM
CPUs. However, in C++, these operations are undefined behavior (UB).
Substituting these operations with a `memcpy` fixes this UB. The
compiled assembly code is equivalent to the original, so there is no
performance penalty to using memcpy.
For context: The unaligned load and store operations fixed here were
originally introduced in the slab allocator (#4218).
Global variables in different TUs are initialized in an undefined order.
At least one global variable was accessing a global switchover variable.
This caused the switchover variable to be accessed in an uninitialized
state.
Since the switchover is always explicitly set before transaction
processing, this bug can not effect transaction processing, but could
effect unit tests (and potentially the value of some global variables).
Note: at the time of this patch the offending bug is not yet in
production.
When instantiating a large amount of fixed-sized objects on the heap
the overhead that dynamic memory allocation APIs impose will quickly
become significant.
In some cases, allocating a large amount of memory at once and using
a slabbing allocator to carve the large block into fixed-sized units
that are used to service requests for memory out will help to reduce
memory fragmentation significantly and, potentially, improve overall
performance.
This commit introduces a new `SlabAllocator<>` class that exposes an
API that is _similar_ to the C++ concept of an `Allocator` but it is
not meant to be a general-purpose allocator.
It should not be used unless profiling and analysis of specific memory
allocation patterns indicates that the additional complexity introduced
will improve the performance of the system overall, and subsequent
profiling proves it.
A helper class, `SlabAllocatorSet<>` simplifies handling of variably
sized objects that benefit from slab allocations.
This commit incorporates improvements suggested by Greg Popovitch
(@greg7mdp).
Commit 1 of 3 in #4218.
Log exception messages at several locations.
Previously, these were locations where an exception was caught, but the
exception message was not logged. Logging the exception messages can be
useful for analysis or debugging. The additional logging could have a
small negative performance impact.
Fix#3213.
Fix a case where `ripple::Expected` returned a json array, not a value.
The problem was that `Expected` invoked the wrong constructor for the
expected type, which resulted in a constructor that took multiple
arguments being interpreted as an array.
A proposed fix was provided by @godexsoft, which involved a minor
adjustment to three constructors that replaces the use of curly braces
with parentheses. This makes `Expected` usable for
[Clio](https://github.com/XRPLF/clio).
A unit test is also included to ensure that the issue doesn't occur
again in the future.
- Implement the `operator==` and the `operator<=>` (aka the spaceship
operator) in `base_uint`, `Issue`, and `Book`.
- C++20-compliant compilers automatically provide the remaining
comparison operators (e.g. `operator<`, `operator<=`, ...).
- Remove the function compare() because it is no longer needed.
- Maintain the same semantics as the existing code.
- Add some unit tests to gain further confidence.
- Fix#2525.
Three static member functions are introduced with
definitions consistent with std::numeric_limits:
static constexpr Number min() noexcept;
Returns: The minimum positive value. This is the value closest to zero.
static constexpr Number max() noexcept;
Returns: The maximum possible value.
static constexpr Number lowest() noexcept;
Returns: The negative value which is less than all other values.
You can set a thread-local flag to direct Number how to round
non-exact results with the syntax:
Number::rounding_mode prev_mode = Number::setround(Number::towards_zero);
This flag will stay in effect for this thread only until another call
to setround. The previously set rounding mode is returned.
You can also retrieve the current rounding mode with:
Number::rounding_mode current_mode = Number::getround();
The available rounding modes are:
* to_nearest : Rounds to nearest representable value. On tie, rounds
to even.
* towards_zero : Rounds towards zero.
* downward : Rounds towards negative infinity.
* upward : Rounds towards positive infinity.
The default rounding mode is to_nearest.
* Conversions to Number are implicit
* Conversions away from Number are explicit and potentially lossy
* If lossy, round to nearest, and to even on tie
* Introduces amendment `XRPFees`
* Convert fee voting and protocol messages to use XRPAmounts
* Includes Validations, Change transactions, the "Fees" ledger object,
and subscription messages
* Improve handling of 0 drop reference fee with TxQ. For use with networks that do not want to require fees
* Note that fee escalation logic is still in place, which may cause the
open ledger fee to rise if the network is busy. 0 drop transactions
will still queue, and fee escalation can be effectively disabled by
modifying the configuration on all nodes
* Change default network reserves to match Mainnet
* Name the new SFields *Drops (not *XRP)
* Reserve SField IDs for Hooks
* Clarify comments explaining the ttFEE transaction field validation
* Improve move semantics in Expected:
This patch unconditionally moves an `Unexpected<U>` value parameter as
long as `U` is not a reference. If `U` is a reference the code should
not compile. An error type that holds a reference is a strange use-case,
and an overload is not provided. If it is required in the future it can
be added.
The `Expected(U r)` overload should take a forwarding ref.
* Replace enable_if with concepts in Expected
When starting, the code generates a new ephemeral private key and
a corresponding certificate to go along with it. This process can
take time and, while this is unlikely to matter for normal server
operations, it can have a significant impact for unit testing and
development. Profiling data suggests that ~20% of the time needed
for a unit test run can be attributed to this.
This commit does several things:
1. It restructures the code so that a new self-signed certificate
and its corresponding private key are only initialized once at
startup; this has minimal impact on the operation of a regular
server.
2. It provides new default DH parameters. This doesn't impact the
security of the connection, but those who compile from scratch
can generate new parameters if they so choose.
3. It properly sets the version number in the certificate, fixing
issue #4007; thanks to @donovanhide for the report.
4. It uses SHA-256 instead of SHA-1 as the hash algorithm for the
certificate and adds some X.509 extensions as well as a random
128-bit serial number.
5. It rounds the certificate's "start of validity" period so that
the server's precise startup time cannot be easily deduced and
limits the validity period to two years, down from ten years.
6. It removes some CBC-based ciphers from the default cipher list
to avoid some potential security issues, such as CVE-2016-2107
and CVE-2013-0169.
The existing spinlock code, used to protect SHAMapInnerNode
child lists, has a mistake that can allow the same child to
be repeatedly locked under some circumstances.
The bug was in the `SpinBitLock::lock` loop condition check
and would result in the loop terminating early.
This commit fixes this and further simplifies the lock loop
making the correctness of the code easier to verify without
sacrificing performance.
It also promotes the spinlock class from an implementation
detail to a more general purpose, easier to use lock class
with clearer semantics. Two different lock types now allow
developers to easily grab either a single spinlock from an
a group of spinlocks (packed in an unsigned integer) or to
grab all of the spinlocks at once.
While this commit makes spinlocks more widely available to
developers, they are rarely the best tool for the job. Use
them judiciously and only after careful consideration.
Adds support to TaggedCache to support smart replacement
(Needed to avoid race conditions with negative caching.)
Create a "hotDUMMY" object that represents the absence
of an object.
Allow DatabaseNodeImp::asyncFetch to complete immediately
if object is in cache (positive or negative).
Fix a bug in asyncFetch where the object returned may not
be the correct canonical version because we stash the
object in the results array before we canonicalize it.
* Abort background path finding when closed or disconnected
* Exit pathfinding job thread if there are no requests left
* Don't bother creating the path find job if there are no requests
* Refactor to remove circular dependency between InfoSub and PathRequest
