Command line arguments

[AntoninoDAnna]

Hi, this is more of a feature request then an issue, unless the feature is already there and I simply didn't look for it hard enough.

How hard would it be to run a script with command line arguments? I don't mean the julia-repl-switches that already allow you to set, for example, -p 4 or --project=my/proj/ but command lines argument that enters in ARGS

The way I see, it should allow to either execute the current buffer with a set of command line arguments as in

julia <julia-repl-switches> my_code.jl args_1 args_2 ...

or, to open a repl while passing to julia some command line arguments.
One idea could be to map the keybinds that currently open a reply to their relative `M-' keybinds as

C-c C-a ->  M-c M-a RET <args...> RET

and should works only if the reply is not open yet.

In this sense, for example, the combo M-c M-b would run as in my first example

julia <julia-repl-switches> my_code.jl args_1 args_2 ...

[tpapp]

In theory I could add this as a feature, but can you clarify the use case a bit?

If it is for testing a script, given that you can always push/append to ARGS, you can replicate any command line arguments easily by either directly invoking something like

julia> append!(ARGS, ["a", "b", "c"])

or sending it to the REPL from Emacs after the process has started.

[AntoninoDAnna]

I think of it as streamlining the process. The way you suggest, I have to

1. open the REPL
2. append/push the keyward arguments to ARGS
3. send the code to REPL from Emacs.

and this would be the most "general" way, in the sense that this process would works with any code. Of course I could just append to ARGS in the code file, effectively merging steps 2 and 3, but then I have to remember to remove those lines before I use the code as intented.

As a use case: I do data analysis and my data set are organized and label with a 4-char string. however, the actual analysis steps are basically the same for each set with small difference that can be handled with Dict{String,Any} and using the id as a key. Now I can write a code that apply to every set and by calling

julia my_code <id>

I can run the code for the set I'm intersted in.

However, the small differences between sets need to be addressed and coded case by case beforehand, so often I use the code itself for a particular set, running it line by line and figuring out the differences as I go. In this process, using the key combinations C-c C-c and C-c C-b is very helpfull, although there is the extra overhead that now I need either to push to ARGS the arguments I need, or simply skip that process and define manually the variable I read from ARGS. If I could do, for example

M-c M-a RET <args>... RET

I would merge step 1 and 2 (open REPL and append to ARGS), making, at least in my opinion, more intuitive and remove the extra step of appending to ARGS manually.

The philosophy would be same as for C-c C-a, you avoid having to open a REPL and activate manually the project in the parent-folders.

[tpapp]

Sorry for not following up, I was busy with work.

What do you think of the following: when julia-repl is called with a prefix argument (eg C-u C-c C-z), it would prompt for arguments, and then pass these to the process it opens. If there is a process already open, it would throw an error (for now).

[AntoninoDAnna]

It seems nice to me. I would change it to a M-c M-z, instead, but that something I can customize on my end, but the general idea sound right to me