Feature #15112: Introducing the short form of `STDERR.puts expr.inspect`.

• Status: Assigned
• Assignee: matz (Yukihiro Matsumoto)
• Target version: -

[mrkn (Kenta Murata)]

I sometimes write STDERR.puts obj.inspect to print debug message to standard error.
I want to write it as STDERR.p obj.
It can be realized by introducing p singleton method in STDERR object.

[duerst (Martin Dürst)]

#warn writes to STDERR. What about warn_p or something similar?

[jeremyevans0 (Jeremy Evans)]

mrkn (Kenta Murata) wrote:

I sometimes write STDERR.puts obj.inspect to print debug message to standard error.
I want to write it as STDERR.p obj.
It can be realized by introducing p singleton method in STDERR object.

I've heard it is preferable to use $stderr (the current standard error stream) instead of STDERR (the original standard error stream) as warn and similar methods use $stderr. As $stderr can be reassigned, there is no guarantee that $stderr.p would be valid. I suppose you could have STDERR.p write to $stderr, but that may be confusing. It may be more generally useful to add a method to IO that did the equivalent of puts obj.inspect.

[shevegen (Robert A. Heiler)]

I agree with both what Kenta Murata and what Jeremy Evans wrote (to explain,
I have no personal preference; I think both .p() would be useful anyway and
I think this was the spirit of the original suggestion; I love p and pp).

Martin suggested warn_p so I think this addresses more Kenta Murata, but I
think, personally, the net benefit of warn_p may be significantly smaller
than the "puts" versus "p" situation, where I think the gain is quite
significant here, if you write a lot of code that makes use of p for
output. (This is a bit similar to the proposal to do "require 'pp'" by
default, which was a good change IMO).

Anyway, my personal opinion is +1 to the idea behind the proposal - I have
no real preference on it being on STDERR or $stderr or both; I think it may
be useful on its own, no matter where the method ultimately resides.

[nobu (Nobuyoshi Nakada)]

Any object which has #write method can be assigned to $stderr, so we should not expect $stderr to have p, I think.

[mrkn (Kenta Murata)]

• Related to Feature #14609: Let `Kernel#p` without an argument print the receiver added

[usa (Usaku NAKAMURA)]

I propose more generalized method on IO instead of STDERR.p.
My proposal is IO#putp.

[mrkn (Kenta Murata)]

• Subject changed from Introdudce the new singleton method STDERR.p to Introducing the short form of `STDERR.puts expr.inspect`.

I totally agree with Jeremy and Usaku, so I want to change the title of this issue.

[hsbt (Hiroshi SHIBATA)]

• Status changed from Open to Assigned