ruby/lib/rubygems/version.rb
Hartley McGuire 9b49ba5a68
[rubygems/rubygems] Speed up Version#<=> ~20-50% when lengths differ
Previously, the comparison code would loop through segments up to the
longest of the two versions being compared. However, this is inefficient
because once one version has more segments than the other we can do a
lot less work.

This commit optimizes the differing segment length case by specializing
the logic once the iteration has passed the shorter of the two segment
lengths. At this point we only need to continue looking at the longer
version's segment, and we know that any String encountered means the
version is less than (pre), and any non-zero Integer means the version
is greater than.

Benchmark:

```
{
  first: [Gem::Version.new("1.2.3"), Gem::Version.new("2.2.3")],
  second: [Gem::Version.new("1.2.3"), Gem::Version.new("1.3.3")],
  third: [Gem::Version.new("1.2.3"), Gem::Version.new("1.2.4")],
  length: [Gem::Version.new("1.2.3"), Gem::Version.new("1.2.3.4")],
  left_s_second: [Gem::Version.new("1.a.3"), Gem::Version.new("1.2.3")],
  left_s_third: [Gem::Version.new("1.2.a"), Gem::Version.new("1.2.3")],
  right_s_second: [Gem::Version.new("1.2.3"), Gem::Version.new("1.a.3")],
  right_s_third: [Gem::Version.new("1.2.3"), Gem::Version.new("1.2.a")],
  left_s_length: [Gem::Version.new("8.0.1.pre"), Gem::Version.new("8.0.1")],
  right_s_length: [Gem::Version.new("8.0.1"), Gem::Version.new("8.0.1.pre")],
  both_s: [Gem::Version.new("8.0.2.pre1"), Gem::Version.new("8.0.2.pre2")],
}.each do |name, v|
  puts "== #{name} =="

  raise name unless v[0].fast_comp(v[1]) == (v[0] <=> v[1])

  Benchmark.ips do |x|
    x.report("fast") { v[0].fast_comp(v[1]) }
    x.report("original") { v[0] <=> v[1] }
    x.compare!(order: :baseline)
  end
end

== first ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   208.555k i/100ms
            original   199.789k i/100ms
Calculating -------------------------------------
                fast      2.075M (± 6.0%) i/s  (481.93 ns/i) -     10.428M in   5.055818s
            original      2.045M (± 3.9%) i/s  (488.94 ns/i) -     10.389M in   5.090034s

Comparison:
                fast:  2075002.8 i/s
            original:  2045227.4 i/s - same-ish: difference falls within error

== second ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   192.395k i/100ms
            original   183.000k i/100ms
Calculating -------------------------------------
                fast      1.892M (± 3.8%) i/s  (528.62 ns/i) -      9.620M in   5.094104s
            original      1.824M (± 3.5%) i/s  (548.11 ns/i) -      9.150M in   5.023163s

Comparison:
                fast:  1891722.2 i/s
            original:  1824435.3 i/s - same-ish: difference falls within error

== third ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   172.788k i/100ms
            original   162.934k i/100ms
Calculating -------------------------------------
                fast      1.719M (± 9.0%) i/s  (581.72 ns/i) -      8.467M in   5.025861s
            original      1.638M (± 3.6%) i/s  (610.36 ns/i) -      8.310M in   5.080344s

Comparison:
                fast:  1719042.9 i/s
            original:  1638389.6 i/s - same-ish: difference falls within error

== length ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   191.741k i/100ms
            original   155.952k i/100ms
Calculating -------------------------------------
                fast      1.920M (± 3.9%) i/s  (520.74 ns/i) -      9.587M in   5.002328s
            original      1.576M (± 6.2%) i/s  (634.42 ns/i) -      7.954M in   5.072507s

Comparison:
                fast:  1920362.1 i/s
            original:  1576240.9 i/s - 1.22x  slower

== left_s_second ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   176.441k i/100ms
            original   164.879k i/100ms
Calculating -------------------------------------
                fast      1.609M (± 7.3%) i/s  (621.51 ns/i) -      8.116M in   5.083414s
            original      1.620M (± 8.3%) i/s  (617.43 ns/i) -      8.079M in   5.028525s

Comparison:
                fast:  1608994.8 i/s
            original:  1619606.5 i/s - same-ish: difference falls within error

== left_s_third ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   160.562k i/100ms
            original   152.799k i/100ms
Calculating -------------------------------------
                fast      1.591M (± 3.6%) i/s  (628.40 ns/i) -      8.028M in   5.052029s
            original      1.528M (± 3.6%) i/s  (654.31 ns/i) -      7.640M in   5.007526s

Comparison:
                fast:  1591334.1 i/s
            original:  1528320.6 i/s - same-ish: difference falls within error

== right_s_second ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   135.938k i/100ms
            original   132.907k i/100ms
Calculating -------------------------------------
                fast      1.367M (± 1.2%) i/s  (731.77 ns/i) -      6.933M in   5.074030s
            original      1.320M (± 2.4%) i/s  (757.35 ns/i) -      6.645M in   5.036155s

Comparison:
                fast:  1366548.7 i/s
            original:  1320386.4 i/s - same-ish: difference falls within error

== right_s_third ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   129.971k i/100ms
            original   123.802k i/100ms
Calculating -------------------------------------
                fast      1.273M (± 4.1%) i/s  (785.25 ns/i) -      6.369M in   5.011805s
            original      1.215M (± 1.8%) i/s  (823.04 ns/i) -      6.190M in   5.096330s

Comparison:
                fast:  1273487.0 i/s
            original:  1215002.9 i/s - same-ish: difference falls within error

== left_s_length ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   211.093k i/100ms
            original   155.784k i/100ms
Calculating -------------------------------------
                fast      2.120M (± 1.9%) i/s  (471.63 ns/i) -     10.766M in   5.079336s
            original      1.565M (± 6.7%) i/s  (638.87 ns/i) -      7.789M in   5.007522s

Comparison:
                fast:  2120296.1 i/s
            original:  1565258.0 i/s - 1.35x  slower

== right_s_length ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   213.977k i/100ms
            original   142.990k i/100ms
Calculating -------------------------------------
                fast      2.154M (± 1.3%) i/s  (464.15 ns/i) -     10.913M in   5.066124s
            original      1.446M (± 1.8%) i/s  (691.75 ns/i) -      7.292M in   5.046172s

Comparison:
                fast:  2154455.3 i/s
            original:  1445607.9 i/s - 1.49x  slower

== both_s ==
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                fast   154.903k i/100ms
            original   131.011k i/100ms
Calculating -------------------------------------
                fast      1.515M (± 4.0%) i/s  (659.97 ns/i) -      7.590M in   5.019890s
            original      1.280M (± 5.3%) i/s  (781.28 ns/i) -      6.420M in   5.035387s

Comparison:
                fast:  1515223.3 i/s
            original:  1279957.8 i/s - 1.18x  slower
```

7195e77152
2025-03-18 09:42:38 +09:00

441 lines
13 KiB
Ruby

# frozen_string_literal: true
require_relative "deprecate"
##
# The Version class processes string versions into comparable
# values. A version string should normally be a series of numbers
# separated by periods. Each part (digits separated by periods) is
# considered its own number, and these are used for sorting. So for
# instance, 3.10 sorts higher than 3.2 because ten is greater than
# two.
#
# If any part contains letters (currently only a-z are supported) then
# that version is considered prerelease. Versions with a prerelease
# part in the Nth part sort less than versions with N-1
# parts. Prerelease parts are sorted alphabetically using the normal
# Ruby string sorting rules. If a prerelease part contains both
# letters and numbers, it will be broken into multiple parts to
# provide expected sort behavior (1.0.a10 becomes 1.0.a.10, and is
# greater than 1.0.a9).
#
# Prereleases sort between real releases (newest to oldest):
#
# 1. 1.0
# 2. 1.0.b1
# 3. 1.0.a.2
# 4. 0.9
#
# If you want to specify a version restriction that includes both prereleases
# and regular releases of the 1.x series this is the best way:
#
# s.add_dependency 'example', '>= 1.0.0.a', '< 2.0.0'
#
# == How Software Changes
#
# Users expect to be able to specify a version constraint that gives them
# some reasonable expectation that new versions of a library will work with
# their software if the version constraint is true, and not work with their
# software if the version constraint is false. In other words, the perfect
# system will accept all compatible versions of the library and reject all
# incompatible versions.
#
# Libraries change in 3 ways (well, more than 3, but stay focused here!).
#
# 1. The change may be an implementation detail only and have no effect on
# the client software.
# 2. The change may add new features, but do so in a way that client software
# written to an earlier version is still compatible.
# 3. The change may change the public interface of the library in such a way
# that old software is no longer compatible.
#
# Some examples are appropriate at this point. Suppose I have a Stack class
# that supports a <tt>push</tt> and a <tt>pop</tt> method.
#
# === Examples of Category 1 changes:
#
# * Switch from an array based implementation to a linked-list based
# implementation.
# * Provide an automatic (and transparent) backing store for large stacks.
#
# === Examples of Category 2 changes might be:
#
# * Add a <tt>depth</tt> method to return the current depth of the stack.
# * Add a <tt>top</tt> method that returns the current top of stack (without
# changing the stack).
# * Change <tt>push</tt> so that it returns the item pushed (previously it
# had no usable return value).
#
# === Examples of Category 3 changes might be:
#
# * Changes <tt>pop</tt> so that it no longer returns a value (you must use
# <tt>top</tt> to get the top of the stack).
# * Rename the methods to <tt>push_item</tt> and <tt>pop_item</tt>.
#
# == RubyGems Rational Versioning
#
# * Versions shall be represented by three non-negative integers, separated
# by periods (e.g. 3.1.4). The first integers is the "major" version
# number, the second integer is the "minor" version number, and the third
# integer is the "build" number.
#
# * A category 1 change (implementation detail) will increment the build
# number.
#
# * A category 2 change (backwards compatible) will increment the minor
# version number and reset the build number.
#
# * A category 3 change (incompatible) will increment the major build number
# and reset the minor and build numbers.
#
# * Any "public" release of a gem should have a different version. Normally
# that means incrementing the build number. This means a developer can
# generate builds all day long, but as soon as they make a public release,
# the version must be updated.
#
# === Examples
#
# Let's work through a project lifecycle using our Stack example from above.
#
# Version 0.0.1:: The initial Stack class is release.
# Version 0.0.2:: Switched to a linked=list implementation because it is
# cooler.
# Version 0.1.0:: Added a <tt>depth</tt> method.
# Version 1.0.0:: Added <tt>top</tt> and made <tt>pop</tt> return nil
# (<tt>pop</tt> used to return the old top item).
# Version 1.1.0:: <tt>push</tt> now returns the value pushed (it used it
# return nil).
# Version 1.1.1:: Fixed a bug in the linked list implementation.
# Version 1.1.2:: Fixed a bug introduced in the last fix.
#
# Client A needs a stack with basic push/pop capability. They write to the
# original interface (no <tt>top</tt>), so their version constraint looks like:
#
# gem 'stack', '>= 0.0'
#
# Essentially, any version is OK with Client A. An incompatible change to
# the library will cause them grief, but they are willing to take the chance
# (we call Client A optimistic).
#
# Client B is just like Client A except for two things: (1) They use the
# <tt>depth</tt> method and (2) they are worried about future
# incompatibilities, so they write their version constraint like this:
#
# gem 'stack', '~> 0.1'
#
# The <tt>depth</tt> method was introduced in version 0.1.0, so that version
# or anything later is fine, as long as the version stays below version 1.0
# where incompatibilities are introduced. We call Client B pessimistic
# because they are worried about incompatible future changes (it is OK to be
# pessimistic!).
#
# == Preventing Version Catastrophe:
#
# From: https://www.zenspider.com/ruby/2008/10/rubygems-how-to-preventing-catastrophe.html
#
# Let's say you're depending on the fnord gem version 2.y.z. If you
# specify your dependency as ">= 2.0.0" then, you're good, right? What
# happens if fnord 3.0 comes out and it isn't backwards compatible
# with 2.y.z? Your stuff will break as a result of using ">=". The
# better route is to specify your dependency with an "approximate" version
# specifier ("~>"). They're a tad confusing, so here is how the dependency
# specifiers work:
#
# Specification From ... To (exclusive)
# ">= 3.0" 3.0 ... &infin;
# "~> 3.0" 3.0 ... 4.0
# "~> 3.0.0" 3.0.0 ... 3.1
# "~> 3.5" 3.5 ... 4.0
# "~> 3.5.0" 3.5.0 ... 3.6
# "~> 3" 3.0 ... 4.0
#
# For the last example, single-digit versions are automatically extended with
# a zero to give a sensible result.
class Gem::Version
include Comparable
VERSION_PATTERN = '[0-9]+(?>\.[0-9a-zA-Z]+)*(-[0-9A-Za-z-]+(\.[0-9A-Za-z-]+)*)?' # :nodoc:
ANCHORED_VERSION_PATTERN = /\A\s*(#{VERSION_PATTERN})?\s*\z/ # :nodoc:
##
# A string representation of this Version.
def version
@version
end
alias_method :to_s, :version
##
# True if the +version+ string matches RubyGems' requirements.
def self.correct?(version)
nil_versions_are_discouraged! if version.nil?
ANCHORED_VERSION_PATTERN.match?(version.to_s)
end
##
# Factory method to create a Version object. Input may be a Version
# or a String. Intended to simplify client code.
#
# ver1 = Version.create('1.3.17') # -> (Version object)
# ver2 = Version.create(ver1) # -> (ver1)
# ver3 = Version.create(nil) # -> nil
def self.create(input)
if self === input # check yourself before you wreck yourself
input
elsif input.nil?
nil_versions_are_discouraged!
nil
else
new input
end
end
@@all = {}
@@bump = {}
@@release = {}
def self.new(version) # :nodoc:
return super unless self == Gem::Version
@@all[version] ||= super
end
def self.nil_versions_are_discouraged!
unless Gem::Deprecate.skip
warn "nil versions are discouraged and will be deprecated in Rubygems 4"
end
end
private_class_method :nil_versions_are_discouraged!
##
# Constructs a Version from the +version+ string. A version string is a
# series of digits or ASCII letters separated by dots.
def initialize(version)
unless self.class.correct?(version)
raise ArgumentError, "Malformed version number string #{version}"
end
# If version is an empty string convert it to 0
version = 0 if version.is_a?(String) && /\A\s*\Z/.match?(version)
@version = version.to_s
# optimization to avoid allocation when given an integer, since we know
# it's to_s won't have any spaces or dashes
unless version.is_a?(Integer)
@version = @version.strip
@version.gsub!("-",".pre.")
end
@version = -@version
@segments = nil
end
##
# Return a new version object where the next to the last revision
# number is one greater (e.g., 5.3.1 => 5.4).
#
# Pre-release (alpha) parts, e.g, 5.3.1.b.2 => 5.4, are ignored.
def bump
@@bump[self] ||= begin
segments = self.segments
segments.pop while segments.any? {|s| String === s }
segments.pop if segments.size > 1
segments[-1] = segments[-1].succ
self.class.new segments.join(".")
end
end
##
# A Version is only eql? to another version if it's specified to the
# same precision. Version "1.0" is not the same as version "1".
def eql?(other)
self.class === other && @version == other.version
end
def hash # :nodoc:
canonical_segments.hash
end
def init_with(coder) # :nodoc:
yaml_initialize coder.tag, coder.map
end
def inspect # :nodoc:
"#<#{self.class} #{version.inspect}>"
end
##
# Dump only the raw version string, not the complete object. It's a
# string for backwards (RubyGems 1.3.5 and earlier) compatibility.
def marshal_dump
[@version]
end
##
# Load custom marshal format. It's a string for backwards (RubyGems
# 1.3.5 and earlier) compatibility.
def marshal_load(array)
string = array[0]
raise TypeError, "wrong version string" unless string.is_a?(String)
initialize string
end
def yaml_initialize(tag, map) # :nodoc:
@version = -map["version"]
@segments = nil
@hash = nil
end
def encode_with(coder) # :nodoc:
coder.add "version", @version
end
##
# A version is considered a prerelease if it contains a letter.
def prerelease?
unless instance_variable_defined? :@prerelease
@prerelease = /[a-zA-Z]/.match?(version)
end
@prerelease
end
def pretty_print(q) # :nodoc:
q.text "Gem::Version.new(#{version.inspect})"
end
##
# The release for this version (e.g. 1.2.0.a -> 1.2.0).
# Non-prerelease versions return themselves.
def release
@@release[self] ||= if prerelease?
segments = self.segments
segments.pop while segments.any? {|s| String === s }
self.class.new segments.join(".")
else
self
end
end
def segments # :nodoc:
_segments.dup
end
##
# A recommended version for use with a ~> Requirement.
def approximate_recommendation
segments = self.segments
segments.pop while segments.any? {|s| String === s }
segments.pop while segments.size > 2
segments.push 0 while segments.size < 2
recommendation = "~> #{segments.join(".")}"
recommendation += ".a" if prerelease?
recommendation
end
##
# Compares this version with +other+ returning -1, 0, or 1 if the
# other version is larger, the same, or smaller than this
# one. Attempts to compare to something that's not a
# <tt>Gem::Version</tt> or a valid version String return +nil+.
def <=>(other)
return self <=> self.class.new(other) if (String === other) && self.class.correct?(other)
return unless Gem::Version === other
return 0 if @version == other.version || canonical_segments == other.canonical_segments
lhsegments = canonical_segments
rhsegments = other.canonical_segments
lhsize = lhsegments.size
rhsize = rhsegments.size
limit = (lhsize > rhsize ? rhsize : lhsize)
i = 0
while i < limit
lhs = lhsegments[i]
rhs = rhsegments[i]
i += 1
next if lhs == rhs
return -1 if String === lhs && Numeric === rhs
return 1 if Numeric === lhs && String === rhs
return lhs <=> rhs
end
lhs = lhsegments[i]
if lhs.nil?
rhs = rhsegments[i]
while i < rhsize
return 1 if String === rhs
return -1 unless rhs.zero?
rhs = rhsegments[i += 1]
end
else
while i < lhsize
return -1 if String === lhs
return 1 unless lhs.zero?
lhs = lhsegments[i += 1]
end
end
0
end
# remove trailing zeros segments before first letter or at the end of the version
def canonical_segments
@canonical_segments ||= begin
# remove trailing 0 segments, using dot or letter as anchor
# may leave a trailing dot which will be ignored by partition_segments
canonical_version = @version.sub(/(?<=[a-zA-Z.])[.0]+\z/, "")
# remove 0 segments before the first letter in a prerelease version
canonical_version.sub!(/(?<=\.|\A)[0.]+(?=[a-zA-Z])/, "") if prerelease?
partition_segments(canonical_version)
end
end
def freeze
prerelease?
_segments
canonical_segments
super
end
protected
def _segments
# segments is lazy so it can pick up version values that come from
# old marshaled versions, which don't go through marshal_load.
# since this version object is cached in @@all, its @segments should be frozen
@segments ||= partition_segments(@version)
end
def partition_segments(ver)
ver.scan(/\d+|[a-z]+/i).map! do |s|
/\A\d/.match?(s) ? s.to_i : -s
end.freeze
end
end