Feature #14033: Add String#append

• Status: Open
• Assignee: -
• Target version: -

[sos4nt (Stefan Schüßler)]

Ruby 2.5 introduces Array#append as an alias for << / push.

Likewise, there should be String#append as an alias for << / concat. The documentation for << even says "append":

Append—Concatenates the given object to str. [...]

And String already has a prepend method.

[shevegen (Robert A. Heiler)]

I +1 agree with Stefan for symmetry. :)

However had, I may be biased since I discussed in favour of the
Array-methods .prepend() and .append() so it may be best if others
can comment here too, to avoid "mono-view bias". :)

[phluid61 (Matthew Kerwin)]

I am opposed to this alias as proposed.

1. << is known to be "dangerous", but other "safe" concatenation operations exist for String (e.g. +). I would expect that a new word-based name for the dangerous operation would end with "!".

2. << (and concat) casts integers as codepoints. prepend requires that all arguments are Stringy (c.f. +). A method named to look like prepend, but that behaves like concat, is confusing.

3. << only accepts a single argument, but concat accepts many, so this alias would introduce confusion about which method (concat vs. append) has what arity.

If the proposal were about adding new functionality I think it would be worth considering, but I'm against adding a new name for the sake of it.

[sos4nt (Stefan Schüßler)]

phluid61 (Matthew Kerwin) wrote:

1 [...] I would expect that a new word-based name for the dangerous operation would end with "!".

That contradicts Array#append / Array#prepend. And since String#prepend also modifies the receiver, I would expect String#append to work in a similar way.

(Off topic but IMO, String#concat is the one that should return a new string, just like the documentation for String#+ says: "Concatenation—Returns a new String containing other_str concatenated to str." But it's probably a bit late to fix that.)

2 [...] A method named to look like prepend, but that behaves like concat, is confusing.

Good point. I'm fine with string-only arguments. append would simply invoke rb_str_append then (is that a coincidence?).

3 [...] this alias would introduce confusion about which method (concat vs. append) has what arity.

It should of course work (arity-wise) like Array#append, i.e. accept multiple argument. I should have said "concat", not "<<" in the question's title, or better yet not call it an alias in the first place.

---

To avoid any further confusion: I'm proposing a new method String#append with the following signature:

append(other_str1, other_str2,...) → str

Append—Append the given strings to str.

a = "hello "
a.append("world", "!") #=> "hello world!"
a                      #=> "hello world!"

See also #concat.

[sos4nt (Stefan Schüßler)]

• Subject changed from Alias String#<< as String#append to Add String#append

[ioquatix (Samuel Williams)]

I would like to propose to extend this method to be binary safe.

If the destination string uses Encoding::BINARY, the append operation should be a memcpy which doesn't affect the receiver's encoding.