The JuliaInXL system is composed of two parts, a Julia package and an Excel Plugin. The Excel plugin itself consists of an XLL add-in.

The Excel plugin is distributed as an installable exe: JuliaProfessional_JuliaInXL_Addin_vW.X.Y.Z.exe, where vW.X.Y.Z is the associated version number of Julia Professional. Running this installer will install the XLL into \[APPDATA]\Roaming\Microsoft\AddIns\. Use of the JuliaInXL plug-in depends on .NET v4.0 or later being available on the system. After installation, the add-in needs to be enabled via either File -> Options -> Add-ins menu entry or the Developer/Add-In ribbon button.

The Julia language package can be installed by doing (on the Julia REPL) Pkg.add("JuliaInXL"), if you have the rights to that repository. However, you will most likely be provided this package as part of your Julia Professional Bundle. This package depends on two public packages, JuliaWebAPI and Reexport.

The primary supported workflow is for interactive development of Julia programs alongside Excel. Once the packages are installed, start Excel, and a Julia REPL. On the Julia repl, type using JuliaInXL to make the environment ready for calling via Excel. After that, define your functions as normal on the REPL. To try an an example, use a demo file included with the package: include("C:\\Users\\<username>\\AppData\\Local\\JuliaPro-W.X.Y.Z\\Julia-W.X.Y\\share\\julia\\site\\vW.X\\JuliaInXL\\test\\sim.jl") where <username> is your current Windows username and W.X.Y.Z are integers representing the current release of Julia Professional. (When using theses examples, you will need the Distributions.jl package installed. Please run Pkg.add("Distributions") on the Julia REPL if you do not have this already installed.)

Once you have the functions that you need to call from Excel, expose them via the process_async call. This will start the server listening to requests from Excel in the background, and return a connection object that can be used for exposing more functions later. The primary arguments to this function is an array of functions to be exposed, and the endpoint on which to listen to messages. The bind=true parameter tells the function to start listening at the endpoint and wait for client connections.

conn = process_async([simulate, simulateTime], "tcp:https://127.0.0.1:9999", bind=true)

There is a corresponding process function with exactly the same API which listens synchronously, and thus doesn't return. This function should be used if the process is started from a command line, rather than via an iteractive REPL.

Once the server is started, julia functions can be called from Excel using the jlcall worksheet function. The first argument to jlcall is a string, which is the name of the function to be called. Subsequent arguments to the jlcall function are passed as parameters to the Julia function being called. These can be constant literals, or cell references. Arrays can be passed via cell ranges.

If the Julia function returns an array (1d or 2d), then use jlcall as an Excel Array function by selecting a range before entering the function, and pressing Shift-Ctrl-Enter to finish.

Functions exposed to Excel should take floats or strings, or their arrays as arguments. In general, it is a good idea to keep the function arguments as loosely typed as possible. Therefore functions should return integers, floats, or strings; or their arrays. However, arrays of dimensions greater than two are not supported.

Note that Excel stores all numbers as 64 bit IEEE floats. Therefore, be aware of the possibility of truncation if returning large, or high precision, numbers.

Dates are passed in from excel as floating point numbers in its internal encoding (fractional days since 1/1/1900 or 1/1/1904). Thus, they are recieved in Julia functions as floats. They can be converted to Julia DateTime values using the xldate function.

Remember that we started the Julia server in async mode. This means that the REPL is available for interactive use when working with Excel. Creating new definitions of functions that are already exposed will replace them, and the new versions will get called when the sheet is next recalculated.

New functions can be added to the listener interface using the connection object saved from the original process_async call, via the register function. The arguments to register are the connection object returned from process_async, and the function name to expose.

register(conn, simulateArray)

Microsoft and Excel are registered trademarks of Microsoft Corporation