Feature #13303: String#any? as !String#empty?

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

[naruse (Yui NARUSE)]

Once I proposed "some container#nonempty?" on #12075, and understand there's Array#any?.

Today I found String doesn't have such method.

[matz (Yukihiro Matsumoto)]

• Status changed from Assigned to Feedback

Use-case?

Matz.

[herwinw (Herwin Quarantainenet)]

and understand there's Array#any?.

This is a misconception, Array#any? does not check if the array is empty, but if there is a true-ish value in the array:

irb(main):001:0> [false, nil].any?
=> false

This is documented by Enumerable#any?: "If the block is not given, Ruby adds an implicit block of { |obj| obj } that will cause any? to return true if at least one of the collection members is not false or nil."

Beside that, I don't think String#any? is a sensible method name to check if a string is non-empty.

[shevegen (Robert A. Heiler)]

Actually the name .nonempty? is easier to understand than .any? in this
context, or non-empty strings.

I think the only problem is that "nonempty" reads very ... strangely.

I can not come up with a good name either though.

.non_empty?

May seem obvious but I am not sure either there since it is quite long.

I think that nonempty? or non_empty? is better than any? in this context
though.

Ignoring the ruby parser, I guess this here would be one of the shortest,
somewhat natural way to query and ask on an object:

object, are you not empty
object not empty?

I guess the most natural ruby way would still be

object.not_empty?
object.non_empty?

Or perhaps we can ask any container/object if it has at least
one entry. :\

object.at_the_least_one_entry?

The last one is a bit awful though - now .non_empty? or .non_empty?
or .not_empty? would look nicer. :)))

[naruse (Yui NARUSE)]

matz (Yukihiro Matsumoto) wrote:

Use-case?

For example https://github.com/ruby/ruby/commit/0256fa889039295742448ad72634fec4ea638684#diff-9199d649a7165e719f0209c3c789f1bbR482

if h and val = h["value"] and val != ""

[duerst (Martin Dürst)]

any? would definitely be the wrong name, because for Arrays, [].any? is always true.

Of the names proposed so far, I think not_empty? looks best. An alternative may be unempty?, but that may sound decidedly unenglish :-). Another may be any_chars?, which would be colloquially correct, but still has the problem that it works differently from a simple any?.

[rosenfeld (Rodrigo Rosenfeld Rosas)]

Maybe String#filled? or String#filled_in? ?

[mame (Yusuke Endoh)]

duerst (Martin Dürst) wrote:

[].any? is always true.

No, it is always false. I agree that String#any? is a bad name for the behavior, anyway.

I think that what Naruse-san really wants is, still, Object#present?.

if h and val = h["value"] and val != ""

if h and h["value"].present?

[stomar (Marcus Stollsteimer)]

mame (Yusuke Endoh) wrote:

if h and h["value"].present?

I think foo.present? semantically should be the same as !foo.nil? ("is the object present?" = "does the object exist?"), which probably wouldn't make much sense as a method on objects other than booleans.

Also:

• String#filled?: is "x" a "full" string? ...
• String#any_chars?: sounds like there also might be e.g. String#any_bytes?

String#not_empty? or String#non_empty? sounds best so far, IMHO (with "not_empty?" maybe easier to remember esp. for non-native speakers, "!" = "not").

[MSP-Greg (Greg L)]

My vote for is for not needed, with second choice of not_empty?.

1. neg - I don't believe it's always necessary to have pairs of logical attributes/properties, as it certainly clutters up the namespace.

2. pos - Ruby already supports if / unless (we'll consider that not common)

3. pos - It makes for somewhat clearer code when a multi-criteria logical statement does not require the ! operator.

Conversely, all coders should immediately consider a ! a not... So, I don't know if it's really needed...

[naruse (Yui NARUSE)]

I want to use this with &.
Therefore String#empty? is not suitable.
It must returns false if it is empty.

Note that String#present? is also no good because ActiveSupport's present? returns false if its all content are space.

[shyouhei (Shyouhei Urabe)]

possible name of this method:

• #present? is NG because that conflicts with ActiveSupport (AS's #present? have different semantics than what is discussed here).
• #empty? is NG because the OP wants to use it in conjunction with &.
• #nonempty? or #non_empty?
• #notempty? or #not_empty?
  • There has never been a core method that starts with "not-"

Any other ideas?

[MSP-Greg (Greg L)]

shyouhei (Shyouhei Urabe) wrote:

Any other ideas?

Assuming we want to stay away from prefixed or concatenated names, I might suggest -

#content?

[MSP-Greg (Greg L)]

After some more thought (and the desire for a method name that could be used with other objects), I think

#each?

might work for many objects, including those that inherit/include Enumerable.

Simply defined, each? returns true if an #each block will be performed at least once. Nothing about the values, just their existence. It's also rather short...

[sawa (Tsuyoshi Sawada)]

I came up with the method name:

solid?

[arimay (yasuhiro arima)]

"#any?" is preferred, since the same method name is harder to forget and less confusing.
If the Array, Hash and String will have new methods, i propose "#having?" as the new method name.

[Dan0042 (Daniel DeLorme)]

Related to #17330 -> non(&:empty?)