FIX API client written in Julia.
julia> Pkg.clone("[email protected]:DatName/FIX.jl.git")This package can:
- inefficiently convert
Dict{Int64, String}to FIX messages (attaching header and tail) - inefficiently parse incoming messages of
Vector{UInt8}intoDict{Int64, String}. - keep track of messaging history
- give you current (confirmed) outstanding orders
- give you position change since the start of
FIXClient - keep track of messaging rates
julia> using FIX
julia> sock = MbedTLS.connect("127.0.0.1", 1498)
julia> client = FIXClient(sock, message_handler, header, ratelimit)where message_handler is a subtype of AbstractMessageHandler, header is a Dict{Int64, String} with keys 8, 49 and 56. ratelimit is an instance of RateLimit exported by FIX.jl to track order sending rates.
julia> FIX.send_message(client, Dict(1 => "a", 2 => "5"))
julia> FIX.getOpenOrders(client)and so on.
Your custom message_handler must implement method onFIXMessage in order to handle incoming messages. For example:
struct EchoFIXHandler <: AbstractMessageHandler
end
import FIX: onFIXMessage
function onFIXMessage(this::EchoFIXHandler, msg)
@info "received fix message" msg
endMessage parsing and converting can be made much faster. But for now it is (Intel® Core™ i3-6100U CPU @ 2.30GHz × 4):
Placing order benchmark:
julia @benchmark fix.placeOrder_fake(client, "buy", "BTC-EUR", 0.0001001, 12000.0)
BenchmarkTools.Trial:
memory estimate: 13.72 KiB
allocs estimate: 216
--------------
minimum time: 16.269 μs (0.00% GC)
median time: 23.513 μs (0.00% GC)
mean time: 40.526 μs (33.47% GC)
maximum time: 25.061 ms (99.50% GC)
--------------
samples: 10000
evals/sample: 1Parsing message benchmark:
julia> (msg, msg_str) = fix.placeOrder(client, "buy", "BTC-EUR", 0.0001001, 12000.0)
julia> b = [UInt8(x) for x in msg_str]
julia> @benchmark FIX.fixparse(b)
BenchmarkTools.Trial:
memory estimate: 15.17 KiB
allocs estimate: 350
--------------
minimum time: 14.758 μs (0.00% GC)
median time: 16.224 μs (0.00% GC)
mean time: 22.086 μs (12.82% GC)
maximum time: 5.826 ms (98.32% GC)
--------------
samples: 10000
evals/sample: 1