Simplify #inspect

[zverok]

I'd like to raise a discussion on the following change:

# before:
Date.new(2010, 1, 3).inspect
# => #<Date: 2010-01-03 ((2455200j,0s,0n),+0s,2299161j)> 

# after:
Date.new(2010, 1, 3).inspect
# => #<Date: 2010-01-03> 

Justification: for most of contexts where Date could be used (web apps, data conversion scripts, experiments and so on), the "((2455200j,0s,0n),+0s,2299161j)" suffix is meaningless, and just obscures the data reading. Imagine the (pretty simple) example:

# before
(1..10).map { |days| Date.today - days }
# => [#<Date: 2019-08-10 ((2458706j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-09 ((2458705j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-08 ((2458704j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-07 ((2458703j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-06 ((2458702j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-05 ((2458701j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-04 ((2458700j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-03 ((2458699j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-02 ((2458698j,0s,0n),+0s,2299161j)>, #<Date: 2019-08-01 ((2458697j,0s,0n),+0s,2299161j)>]

# after:
(1..10).map { |days| Date.today - days }
 => [#<Date: 2019-08-10>, #<Date: 2019-08-09>, #<Date: 2019-08-08>, #<Date: 2019-08-07>, #<Date: 2019-08-06>, #<Date: 2019-08-05>, #<Date: 2019-08-04>, #<Date: 2019-08-03>, #<Date: 2019-08-02>, #<Date: 2019-08-01>]

Whether it is a list of dates we extract from DB, or have in test (of some scheduling library), or parse from CSV, I believe the "before" example is very hard to read and provides no additional value.

On the other hand, for the rare contexts when values like "julian day number" matter, the ((2458697j,0s,0n),+0s,2299161j) format seem to be too terse to be easily used by anybody not 200% familiar with library's internals.

[jeremyevans]

I agree this makes more sense and will make the inspect output easier to read. I plan to merge this before 2.7 is released unless there are objections.

[nurse]

I object this change.
inspect should suggest its internal and it should keep julian day number.

If you want to introduce such Web friendly date library, don't build it on top of current date library.
Please create new one on top of Time.

[jeremyevans]

@nurse I can revert this, however, first please consider that "inspect should suggest its internal" is no longer a reason to do so. The date library since 1.9.3 does not store the julian day number for dates in the typical case where it is initialized via Date.new/Date.civil or Date.today. If, after considering that, you would still like this reverted, please let me know and I will revert this.

[nurse]

@jeremyevans The fact it doesn't calculate julian day number is implementation detail.
Of course you have a question, what is the difference between "internal" and "implementation detail".

I think Date#inspect should show the difference between Date.today(Date::ITALY) and Date.today(Date::ENGLAND).

# master
irb(main):002:0> Date.today(Date::ITALY)
=> #<Date: 2019-10-27>
irb(main):003:0> Date.today(Date::ENGLAND)
=> #<Date: 2019-10-27>

# 2.6
irb(main):002:0> Date.today(Date::ITALY)
=> #<Date: 2019-10-27 ((2458784j,0s,0n),+0s,2299161j)>
irb(main):003:0> Date.today(Date::ENGLAND)
=> #<Date: 2019-10-27 ((2458784j,0s,0n),+0s,2361222j)>

[jeremyevans]

@nurse OK. I'll revert this change.

[jeremyevans]

Revert committed at 875d563.