可以将文章内容翻译成中文,广告屏蔽插件可能会导致该功能失效(如失效,请关闭广告屏蔽插件后再试):
问题:
Despite reading "Understanding Ruby Symbols", I'm still confused by the representation of the data in memory when it comes to using symbols. If a symbol, two of them contained in different objects, exist in the same memory location, then how is it that they contain different values? I'd have expected the same memory location to contain the same value.
This a quote from the link:
Unlike strings, symbols of the same name are initialized and exist in memory only once during a session of ruby
I don't understand how it manages to differentiate the values contained in the same memory location.
Consider this example:
patient1 = { :ruby => "red" }
patient2 = { :ruby => "programming" }
patient1.each_key {|key| puts key.object_id.to_s}
3918094
patient2.each_key {|key| puts key.object_id.to_s}
3918094
patient1
and patient2
are both hashes, that's fine. :ruby
however is a symbol. If we were to output the following:
patient1.each_key {|key| puts key.to_s}
Then what will be output? "red"
, or "programming"
?
Forgetting hashes for a second, I'm thinking a symbol is a pointer to a value. The questions I have are:
- Can I assign a value to a symbol?
- Is a symbol just a pointer to a variable with a value in it?
- If symbols are global, does that mean a symbol always points to one thing?
回答1:
Consider this:
x = :sym
y = :sym
(x.__id__ == y.__id__ ) && ( :sym.__id__ == x.__id__) # => true
x = "string"
y = "string"
(x.__id__ == y.__id__ ) || ( "string".__id__ == x.__id__) # => false
So, however you create a symbol object, as long as its contents are the same, it will refer to the same object in memory. This is not a problem because a symbol is an immutable object. Strings are mutable.
(In response to the comment below)
In the original article, the value is not being stored in a symbol, it is being stored in a hash. Consider this:
hash1 = { "string" => "value"}
hash2 = { "string" => "value"}
This creates six objects in the memory -- four string objects and two hash objects.
hash1 = { :symbol => "value"}
hash2 = { :symbol => "value"}
This only creates five objects in memory -- one symbol, two strings and two hash objects.
回答2:
I was able to grock symbols when I thought of it like this. A Ruby string is an object that has a bunch of methods and properties. People like to use strings for keys, and when the string is used for a key then all those extra methods aren't used. So they made symbols, which are string objects with all the functionality removed, except that which is needed for it to be a good key.
Just think of symbols as constant strings.
回答3:
The symbol :ruby
does not contain "red"
or "programming"
. The symbol :ruby
is just the symbol :ruby
. It is your hashes, patient1
and patient2
that each contain those values, in each case pointed to by the same key.
Think about it this way: If you go into the living room on christmas morning, and see two boxes with a tag on them that say "Kezzer" on them. On has socks in it, and the other has coal. You're not going to get confused and ask how "Kezzer" can contain both socks and coal, even though it is the same name. Because the name isn't containing the (crappy) presents. It's just pointing at them. Similarly, :ruby
doesn't contain the values in your hash, it just points at them.
回答4:
You might be presuming that the declaration you've made defines the value of a Symbol to be something other than what it is. In fact, a Symbol is just an "internalized" String value that remains constant. It is because they are stored using a simple integer identifier that they are frequently used as that is more efficient than managing a large number of variable-length strings.
Take the case of your example:
patient1 = { :ruby => "red" }
This should be read as: "declare a variable patient1 and define it to be a Hash, and in this store the value 'red' under the key (symbol 'ruby')"
Another way of writing this is:
patient1 = Hash.new
patient1[:ruby] = 'red'
puts patient1[:ruby]
# 'red'
As you are making an assignment it is hardly surprising that the result you get back is identical to what you assigned it with in the first place.
The Symbol concept can be a little confusing as it's not a feature of most other languages.
Each String object is distinct even if the values are identical:
[ "foo", "foo", "foo", "bar", "bar", "bar" ].each do |v|
puts v.inspect + ' ' + v.object_id.to_s
end
# "foo" 2148099960
# "foo" 2148099940
# "foo" 2148099920
# "bar" 2148099900
# "bar" 2148099880
# "bar" 2148099860
Every Symbol with the same value refers to the same object:
[ :foo, :foo, :foo, :bar, :bar, :bar ].each do |v|
puts v.inspect + ' ' + v.object_id.to_s
end
# :foo 228508
# :foo 228508
# :foo 228508
# :bar 228668
# :bar 228668
# :bar 228668
Converting strings to symbols maps identical values to the same unique Symbol:
[ "foo", "foo", "foo", "bar", "bar", "bar" ].each do |v|
v = v.to_sym
puts v.inspect + ' ' + v.object_id.to_s
end
# :foo 228508
# :foo 228508
# :foo 228508
# :bar 228668
# :bar 228668
# :bar 228668
Likewise, converting from Symbol to String creates a distinct string every time:
[ :foo, :foo, :foo, :bar, :bar, :bar ].each do |v|
v = v.to_s
puts v.inspect + ' ' + v.object_id.to_s
end
# "foo" 2148097820
# "foo" 2148097700
# "foo" 2148097580
# "bar" 2148097460
# "bar" 2148097340
# "bar" 2148097220
You can think of Symbol values as being drawn from an internal Hash table and you can see all values that have been encoded to Symbols using a simple method call:
Symbol.all_values
# => [:RUBY_PATCHLEVEL, :vi_editing_mode, :Separator, :TkLSHFT, :one?, :setuid?, :auto_indent_mode, :setregid, :back, :Fail, :RET, :member?, :TkOp, :AP_NAME, :readbyte, :suspend_context, :oct, :store, :WNOHANG, :@seek, :autoload, :rest, :IN_INPUT, :close_read, :type, :filename_quote_characters=, ...
As you define new symbols either by the colon-notation or by using .to_sym this table will grow.
回答5:
Symbols are not pointers. They do not contain values. Symbols simply are. :ruby
is the symbol :ruby
and that's all there is to it. It doesn't contain a value, it doesn't do anything, it just exists as the symbol :ruby
. The symbol :ruby
is a value just like the number 1 is. It doesn't point to another value any more than the number 1 does.
回答6:
patient1.each_key {|key| puts key.to_s}
Then what will be output? "red", or
"programming"?
Neither, it will output "ruby".
You're confusing symbols and hashes. They aren't related, but they're useful together. The symbol in question is :ruby
; it has nothing to do with the values in the hash, and it's internal integer representation will always be the same, and it's "value" (when converted to a string) will always be "ruby".
回答7:
In short
Symbols solve the problem of creating human readable, immutable representations that also have the benefit of being simpler for the runtime to lookup than strings. Think of it like a name or label that can be reused.
Why :red is better than "red"
In dynamic object oriented languages you create complex, nested data structures with readable references. The hash is a common use case where you map values to unique keys — unique, at least, to each instance. You can't have more than one "red" key per hash.
However it would be more processor efficient to use a numeric index instead of string keys. So symbols were introduced as a compromise between speed and readability. Symbols resolve much easier than the equivalent string. By being human readable and easy for the runtime to resolve symbols are an ideal addition to a dynamic language.
Benefits
Since symbols are immutable they can be shared across the runtime. If two hash instances have a common lexicographic or semantic need for a red item the symbol :red would use roughly half the memory that the string "red" would have required for two hashes.
Since :red always resolves back to the same location in memory it can be reused across a hundred hash instances with almost no increase in memory, whereas using "red" will add a memory cost since each hash instance would need to store the mutable string upon creation.
Not sure how Ruby actually implements symbols/string but clearly a symbol offers less implementation overhead in the runtime since it's a fixed representation. Plus symbols takes one less character to type than a quoted string and less typing is the eternal pursuit of true Rubyists.
Summary
With a symbol like :red you get the readability of string representation with less overhead due to the cost of string comparison operations and the need to store each string instance in memory.
回答8:
patient1 = { :ruby => "red" }
patient2 = { :ruby => "programming" }
patient1.each_key {|key| puts key.object_id.to_s}
3918094
patient2.each_key {|key| puts key.object_id.to_s}
3918094
patient1
and patient2
are both hashes, that's fine. :ruby
however is a symbol. If we were to output the following:
patient1.each_key {|key| puts key.to_s}
Then what will be output? "red", or "programming"?
Neither, of course. The output will be ruby
. Which, BTW, you could have found out in less time than it took you to type the question, by simply typing it into IRB instead.
Why would it be red
or programming
? Symbols always evaluate to themselves. The value of the symbol :ruby
is the symbol :ruby
itself and the string representation of the symbol :ruby
is the string value "ruby"
.
[BTW: puts
always converts its arguments to strings, anyway. There's no need to call to_s
on it.]
回答9:
I would recommend reading the Wikipedia article on hash tables - I think it will help you get a sense of what {:ruby => "red"}
really means.
Another exercise that might help your understanding of the situation: consider {1 => "red"}
. Semantically, this doesn't mean "set the value of 1
to "red"
", which is impossible in Ruby. Rather, it means "create a Hash object, and store the value "red"
for the key 1
.
回答10:
I'm new to Ruby, but I think (hope?) this is a simple way to look at it...
A symbol is not a variable or a constant. It doesn't stand in for, or point to, a value. A symbol IS a value.
All it is, is a string without the object overhead. The text and only the text.
So, this:
"hellobuddy"
Is the same as this:
:hellobuddy
Except you can't do, for example, :hellobuddy.upcase. It's the string value and ONLY the string value.
Likewise, this:
greeting =>"hellobuddy"
Is the same as this:
greeting => :hellobuddy
But, again, without the string object overhead.
回答11:
The "pointing" is governed by the hash, not by the symbol.
A symbol is it's own unique thing, like an instance object. or any other data type.
(In fact, you can think of symbols as string constants, whereas a string comprised of the same characters as a symbol would be an instance object of the String class. The benefit is reducing the number of objects created, less memory usage, reduction of funky behavior due to referencing different objects rather than the same symbol object).
using :ruby
, "red"
, "programming"
, patient1
, and patient2
The hash patient1
looks at the symbol :ruby
, then goes, "oh, that corresponds to "red"
The hash patient2
looks at the symbol :ruby
, then goes, "oh, that corresponds to "programming"