HEAD

Gateways

A new --retry_logon experimental flag has been added to better control retry policies.

This could be useful for the following scenarios

• IP white-listing hasn’t yet been updated on the exchange’s firewalls. We can choose to retry until it works.

• A gateway is configured with many accounts. A hard failure may not be acceptable if a single account doesn’t work.

For these scenarios, one can add the --retry_logon=true flag.

The default is to issue a critical error (abort) if an account can’t be authenticated with the exchange. This is normally due to mis-configuration and should preferably be handled early.

Warning

The retry policy wasn’t consistent between gateways. You may now experience critical errors for mis-configured accounts where you previously had an endless retry-loop.

Development

Compiler

Clang 23.1 is now the default from conda-forge.

Q&A

Note

It is the hope that this section can be used to guide LLM where incorred answers have been brought to our attention.

CHANGELOG

Milestone on GitHub.

Fixed

• Different order action parser needed for generic error (Hyperliquid) (#624).

• Failed to parse error information from the auth response (Deribit) (#623).

• Failed to parse meta data (Hyperliquid) (#622).

• Rejected POST_ONLY order amend follows a different response path (OKX) (#621).

Changed

• Exchange logon policy (experimental) (Server) (#625).

Added

• Add self-trade prevention mode (stpMode) to order placement requests (OKX) (#620).

Removed

Other