Can we have print_color and println_color?

[IainNZ]

Title says it all really. We currently have just print_with_color which is verbose and almost every time I've used it I've wanted to have a new line. Hence my "request". I will even invest the 10 hours required myself to make the PR ;)

[dpsanders]

How about a color=true keyword argument for the print and println functions? Or is that a bad idea performance-wise?

[dpsanders]

Sorry, ignore my uninformed comment...

[JeffBezanson]

It seems like a perfectly reasonable comment to me! (am I missing something?)

Generally underscores are unsatisfying. print_color should ideally be separated into two concepts: print and color. A keyword argument would be one way to do that.

[IainNZ]

Keyword argument sounds good too. I'll make a PR in that style

[dpsanders]

Great. Rather than color=true, I guess it would be rather color=:blue

Any chance we could also allow the spelling colour...? ;)

[ivarne]

@dpsanders What would happen if you specify both?

[TotalVerb]

Base.println_with_color exists and just needs to be exported. I see there is some concern with underscore names though, so perhaps this is not desired?

[TotalVerb]

I don't think that a keyword argument would be practical. Since there's no dispatch on keyword arguments, every reimplementation of print would need to accept this keyword argument (correct me if I'm mistaken). Also, print is used often in functions like string, where this additional complexity isn't needed at all. What about these other kinds of syntax?

println(with_color(:blue, "Hello World!))

(with_color constructs an object that overloads print)

with_color(println, :blue, "Hello World!")

(with_color calls the given IO function, consistent with sprint and similar, and with function first so that do can be used—this one already basically exists and just needs to be exported)

This second syntax allows do blocks, which can be useful syntax. Something like this already works:

Base.with_output_color(:blue, STDOUT) do io
     println(io, "Hello World")
end

[IainNZ]

@TotalVerb I like your thinking there! @vtjnash has been thinking about passing that type of information with higher-level IO types as well, I can't find the PR though

[vtjnash]

yep. this is another issue that #13825 would solve (and is very similar to what @TotalVerb described)

[JeffBezanson]

Now Implemented as printstyled.