print() for tuple should use print() to stringify its contents

[mkyt]

When printing tuple, its contents are currently stringified by show() instead of print().

julia> a::Uint16 = 0x4321
0x4321
julia> print(a)
17185
julia> print((a, ))
(0x4321,)

I think it is preferable to use print for displaying its contents when printing tuple
as print and show give different results for some types, as in the example above.

[JeffBezanson]

The reason we do this is that numbers have canonical printed representations, but tuples do not. As soon as we're printing ( ,), we're using julia syntax and might as well stick with it. If we printed (17185,), that would be a representation of a julia object, but of a different julia object, which seems like a strange thing to do.

[kmsquire]

This suggests (again) a python-like distinction between repr and string.

[StefanKarpinski]

It's is a bit messier than ideal right now, but there is a somewhat consistent distinction made already.

[mkyt]

I cannot grasp what "canonical" means in this context, but as I read the discussion in #3450,
I thought show/repr and print/string in julia roughly correspond to python's repr (unambiguous) and str (human-friendly) and so it is more natural and consistent that these distinction propagate to their contents when showing or printing container objects (like tuples).

[JeffBezanson]

By "canonical" I mean not julia-specific. I don't really see the value of printing (17185,) here. Who/what is the consumer for that output?

[mkyt]

Hex notation is quite annoying when debugging codes dealing with tuples containing unsigned ints.

julia> ui, ui_t = uint(123), tuple(uint(456), uint(789))
(0x000000000000007b,(0x00000000000001c8,0x0000000000000315))
julia> show(ui)
0x000000000000007b
julia> print(ui)
123
julia> show(ui_t)
(0x00000000000001c8,0x0000000000000315)
julia> print(ui_t)
(0x00000000000001c8,0x0000000000000315) # (456,789) is much more comprehensible

[jiahao]

Relevant: #7959

[nalimilan]

Even if Julia syntax is used for printing tuples, it doesn't mean the output should be parseable to the same Julia object, if the goal is to offer a short/readable representation (the intended consumer of print is a human).

[johnmyleswhite]

-1 to printing unsigned ints as integers rather than hex

[nalimilan]

@johnmyleswhite I think the debate here is: why would print((0x1,)) use hex while print(0x1) wouldn't, and vice-versa? I.e. a consistency issue.

[StefanKarpinski]

Let's just print everything in hex – consistency problem solved.

[JeffBezanson]

I can totally understand not wanting to see hex in many cases, but I think that's the only real issue here. Consider another example of the general proposal described by the issue title:

julia> print("hello, world.")
hello, world.
julia> print(("hello, world.",))
("hello, world.",)

With the proposed change, the second output would be (hello, world.,). Is that what you want?

[mkyt]

@JeffBezanson, thanks for pointing this out. I wasn't aware of the pitfall.

On second thought, I realized the current behavior of print is "consistent" in that
it uses canonical notation if any, uses repr otherwise, as @StefanKarpinski stated
and it would be better to leave it as it is.

Nevertheless, I think its' very handy to have a easy way of controlling over the level
of detail in printing julia objects out in the planned clean-up of output functions (#7959).