Errata

mgsub doesn't seem like grouping with ():

ex.

"red white blue".gsub(/(white)/,'x\1x')
=> red xwhitex blue

"red white blue".mgsub([[/(white)/,'x\1x']])
=> red x\\1x blue

There should be a space between "wrong" and the period that follows it

s.split(//).reverse!.join(' ') # => "These words are in the wrong . order"

There should be a space between "wrong" and the period that follows it

s.split(//).reverse!.join(' ') # => "These words are in the wrong . order"

Paragraph refers to chapter 11.2 for details on the use of the iconv library.
Nevertheless iconv is not mentioned in chapter 11.2.

Paragraph refers to chapter 11.2 for details on the use of the iconv library.
Nevertheless iconv is not mentioned in chapter 11.2.

mgsub doesn't seem like grouping with ():

ex.

"red white blue".gsub(/(white)/,'x1x')
=> red xwhitex blue

"red white blue".mgsub([[/(white)/,'x1x']])
=> red x\1x blue

Change
:to_f => /([+-]?([0-9]*\.)?[0-9]+(e[+-]?[0-9]+)?)/i,

to

:to_f => /([+-]?(?:[0-9]*\.)?[0-9]+(?:e[+-]?[0-9]+)?)/i,


The addition of the ?: in the second and third group prevents them from being capture groups so repeating the body of the regular expression multiple times such as

/([+-]?(?:[0-9]*\.)?[0-9]+(?:e[+-]?[0-9]+)?) ([+-]?(?:[0-9]*\.)?[0-9]+(?:e[+-]?[0-9]+)?)/i

to parse two floating point numbers will return the float strings in $1 and $2 as expected, otherwise without the ?: will return the floats in $1 and $3 - much less intuitive.

In the book, the formula to convert a logarithm from base b1 to base b2 looks like this:

log_b1(x) = log_b2(x) / log_b2(k)

however, the 'k' should be 'b1' as in

log_b1(x) = log_b2(x) / log_b2(b1)

See also http://en.wikipedia.org/wiki/Logarithm#Change_of_base

In the book, the formula to convert a logarithm from base b1 to base b2 looks like this:

log_b1(x) = log_b2(x) / log_b2(k)

however, the 'k' should be 'b1' as in

log_b1(x) = log_b2(x) / log_b2(b1)

See also http://en.wikipedia.org/wiki/Logarithm#Change_of_base

On the next to last line on the page, in the line of code that's part of the rescue statement
convert_to = DateTime
is superfluous. The variable convert_to is never referenced.

On the next to last line on the page, in the line of code that's part of the rescue statement
convert_to = DateTime
is superfluous. The variable convert_to is never referenced.

The following code in the two "else" clauses that contain the comment "# Not supported. Delegate to superclass; will probably give an error." is incorrect:

else
# Not supported...
super
sort!(&sort_by)
end
else
# Not supported...
super
sort!(&sort_by)
end

should be:

else
# Not supported...
super
sort!(&@sort_by)
end
else
# Not supported...
super
sort!(&@sort_by)
end

This is no big deal since the code should never execute.

The following code in the two "else" clauses that contain the comment "# Not supported. Delegate to superclass; will probably give an error." is incorrect:

else
# Not supported...
super
sort!(&sort_by)
end
else
# Not supported...
super
sort!(&sort_by)
end

should be:

else
# Not supported...
super
sort!(&@sort_by)
end
else
# Not supported...
super
sort!(&@sort_by)
end

This is no big deal since the code should never execute.

stripes = SortedArray.new(["aardwolf", "zebrafish"])
should be
stripes = FrozenCopySortedArray.new(["aardwolf", "zebrafish"])

then if you ran
stripes[1].upcase!

you will get the following error message:
`upcase!': can't modify frozen string (TypeError)

stripes = SortedArray.new(["aardwolf", "zebrafish"])
should be
stripes = FrozenCopySortedArray.new(["aardwolf", "zebrafish"])

then if you ran
stripes[1].upcase!

you will get the following error message:
`upcase!': can't modify frozen string (TypeError)

The Array#shuffle! method implements the Fisher-Yates shuffle algorithm (Knuth _The Art of Computer Programming_ vol. 2 pg. 145). The implementation works, yet it has a small superfluous step. Namely, we do not need to iterate over "each" index. Once j is equal to the final value (n-1) there is only 1 array element left, and there is no point swapping it with itself.

Here is a version that closely follows Knuth's description (although applied to 0-based indexing):
class Array
def shuffle!
(length-1).downto(1) do |i|
j = rand(i + 1)
self[j], self[i] = self[i], self[j]
end
self
end
end

And here is a slight variant that avoids unnecessary swaps:
class Array
def shuffle!
(length-1).downto(1) do |i|
j = rand(i + 1)
self[j], self[i] = self[i], self[j] unless i == j
end
self
end
end

There's no method called Array#compress!. The correct method for removing nil values is Array#compact!

There's no method called Array#compress!. The correct method for removing nil values is Array#compact!

The following code is incorrect:
array = [1, 2, 3]
set = [3, 4, 5].to_s
array & set
set & array

It should be the following:
array = [1, 2, 3]
set = [3, 4, 5].to_set
array & set
set & array

NOTE: to_s should be to_set

The following code is incorrect:
array = [1, 2, 3]
set = [3, 4, 5].to_s
array & set
set & array

It should be the following:
array = [1, 2, 3]
set = [3, 4, 5].to_set
array & set
set & array

NOTE: to_s should be to_set

The result for the operation to divide the set into sets of adjacent numbers is wrong. There was no element "-3" in Set s. There was a "-2".

The incorrect results in the book are:

#<Set: {#<Set: {1, 2, 3}>,
#<Set: {-1}>,
#<Set: {-4, -3}>}>

The correct results should be:

#<Set: {#<Set: {1, 2, 3}>,
#<Set: {-4}>,
#<Set: {-2, -1}>}>

The following assignment statement should precede the code that uses Set#classify;
s = Set.new([1, 2, 3, 'a', 'b', 'c', -1.0, -2.0, -3.0])

The result for the operation to divide the set into sets of adjacent numbers is wrong. There was no element "-3" in Set s. There was a "-2".

The incorrect results in the book are:

#<Set: {#<Set: {1, 2, 3}>,
#<Set: {-1}>,
#<Set: {-4, -3}>}>

The correct results should be:

#<Set: {#<Set: {1, 2, 3}>,
#<Set: {-4}>,
#<Set: {-2, -1}>}>

The following assignment statement should precede the code that uses Set#classify;
s = Set.new([1, 2, 3, 'a', 'b', 'c', -1.0, -2.0, -3.0])

The into_hash method seems like an
'anti recipe' because one can just use

results = Hash[*squares]

To convert an array with key-value pairs into a hash

d.reject { |f| f[0] == '.' } generates the following output:

[".", "..", ".hidden_file", "ruby_script.rb", "subdirectory", "text_file"]

The problem is that f[0] returns a Fixnum, not a string.

Any of the following will work properly:

d.reject { |f| f[0] == ?. }
d.reject { |f| f[0..0] == '.' }
d.reject { |f| f[/^./] == '.' }
d.reject { |f| f =~ [/^./ }

d.reject { |f| f[0] == '.' } generates the following output:

[".", "..", ".hidden_file", "ruby_script.rb", "subdirectory", "text_file"]

The problem is that f[0] returns a Fixnum, not a string.

Any of the following will work properly:

d.reject { |f| f[0] == ?. }
d.reject { |f| f[0..0] == '.' }
d.reject { |f| f[/^./] == '.' }
d.reject { |f| f =~ [/^./ }

Here's the offending recipe:

#
# original recipe
#
open( 'output', 'w' ) do |f|
flock(f, File::LOCK_EX) do |f|
f << "Kiss me, I've got a write lock on a file!"
end
end

The recipe truncates the file before obtaining the lock. Some other process could have LOCK_EX and be writing into
the file. If the reader realizes this can happen, he'll wonder if the any of the recipe is correct.

I solve this problem by creating a zero-length companion file which takes such rough treatment without ill effect:

#
# suggested changed recipe
#
open( 'output.lck', 'a' ) do |f|
flock( f, File::LOCK_EX)
open( 'output', 'w' ) do |g|
g << "Kiss me, I've got the write lock for this file!"
end
end

How this works:
1. Open an append-write handle on the companion, which creates the companion if it does not already exist.
Derive companion name from the file you really want to write.
2. Request and wait for an exclusive lock on the companion
3. Open a write handle on the actual file.
4. Write to the actual file
5. Close the write handle on the actual file
6. Close the append handle on the companion file. This releases the lock.

General comments:

1. I like cookbook recipes which don't strain credulity. Using the same block variable |f| in two nested
blocks works, but distracts the reader.

2. open(...) do |h| ... end passes the handle created by open() into the block. This is idiosyncratic to Ruby and should be noted.

3. The fact that leaving the block closes the handle is idiosyncratic to Ruby and should be noted.

4. It is not widely appreciated that closing a handle on a file terminates any flock lock you may have on the file through that handle. This is a feature of the OS and not Ruby. But if we note this, reader will completely understand how this recipe works and believe it. He will appreciate a bit of how Ruby is set up for brevity.

Throughout the examples in 6.17, you add "b" to the file mode which is needed when you open a binary file on Windows. In the first set line of code after Discussion, you omit this.

Change
f = open('binary')
to
f = open('binary', 'rb')

Throughout the examples in 6.17, you add "b" to the file mode which is needed when you open a binary file on Windows. In the first set line of code after Discussion, you omit this.

Change
f = open('binary')
to
f = open('binary', 'rb')

If you want to pick numbers between min and max, inclusive, then you should change

limit.times { yield min+rand(max+1) }
to
limit.times { yield min+rand(max) }

The implementation defined in the book chooses numbers between 1 and 50 (rand(49) returns a value between 0 and 48)

The code still has the problem that it can pick the same number more than once.

If you want to pick numbers between min and max, inclusive, then you should change

limit.times { yield min+rand(max+1) }
to
limit.times { yield min+rand(max) }

The implementation defined in the book chooses numbers between 1 and 50 (rand(49) returns a value between 0 and 48)

The code still has the problem that it can pick the same number more than once.

ArgumentException should be ArgumentError (or ArgumentError exception)

ArgumentException should be ArgumentError (or ArgumentError exception)

There are two problems with the code:

1. The line below check = abs should be changed from:
if to_check == 11 or to_check == 12
to
if check == 11 or check == 12 or check == 13

2. There's a line missing from the case expression.
Change from:

case check % 10
when 1 then suffix = "st"
when 2 then suffix = "nd"
else suffix = "th"
end

to:

case check % 10
when 1 then suffix = "st"
when 2 then suffix = "nd"
when 3 then suffix = "rd"
else suffix = "th"
end

You might also want to add to the bottom of the code:

OrdinalNumber.new(3).to_s # => "3rd"
OrdinalNumber.new(13).to_s # => "13th"
OrdinalNumber.new(123).to_s # => "123rd"

There are two problems with the code:

1. The line below check = abs should be changed from:
if to_check == 11 or to_check == 12
to
if check == 11 or check == 12 or check == 13

2. There's a line missing from the case expression.
Change from:

case check % 10
when 1 then suffix = "st"
when 2 then suffix = "nd"
else suffix = "th"
end

to:

case check % 10
when 1 then suffix = "st"
when 2 then suffix = "nd"
when 3 then suffix = "rd"
else suffix = "th"
end

You might also want to add to the bottom of the code:

OrdinalNumber.new(3).to_s # => "3rd"
OrdinalNumber.new(13).to_s # => "13th"
OrdinalNumber.new(123).to_s # => "123rd"

change
a = AppendOnlyArray
to
a = AppendOnlyArray.new

change
a = AppendOnlyArray
to
a = AppendOnlyArray.new

5th coding example reads

API_KEY = "100f7vo4gg".freeze

API_KEY[0] = 4
# TypeError: can't modify frozen string

API_KEY = "400f7vo4gg"
# warning: already initialized constant API_KEY

From the text, the reader is lead to conclude that the warning "already initialized constant" results from the fact that API_KEY has been frozen in the first line.

This is not correct (at least not in Ruby 1.8.6). The warning would be issued even if API_KEY would not have been assigned a frozen string before. It comes from the fact that the variable name starts with an uppercase letter (and hence designates what Ruby calls a "constant"). Indeed, if we would have used variables starting with a lower case letter, i.e.

api_key="....".freeze
api_key[0]=4
api_key="...."

We would not get any warning (although of course, we would still get the type error in the second line).

String does not include Enumerable as the book claims it does.

[29] pry(main)> "asdf".is_a? Enumerable
=> false
[30] pry(main)> String.is_a? Enumerable
=> false

The last line of code before the final end is:
RubyString.new("This is a built-in string, not a StringTheory2::String")

This code should be moved below the end, and changed to:
StringTheory2::RubyString.new("This is a built-in string, not a StringTheory2::String")

The full example is:

module StringTheory2
RubyString = String
class String
def initialize(length=1.0e-33)
@length = length
end
end
end

StringTheory2::RubyString.new("This is a built-in string, not a StringTheory2::String")
#=>"This is a built-in string, not a StringTheory2::String"

The last line of code before the final end is:
RubyString.new("This is a built-in string, not a StringTheory2::String")

This code should be moved below the end, and changed to:
StringTheory2::RubyString.new("This is a built-in string, not a StringTheory2::String")

The full example is:

module StringTheory2
RubyString = String
class String
def initialize(length=1.0e-33)
@length = length
end
end
end

StringTheory2::RubyString.new("This is a built-in string, not a StringTheory2::String")
#=>"This is a built-in string, not a StringTheory2::String"

Although it doesn't change anything, I think the 5th line of code at the top of the page should be changed from:

Welcomer.method("an_instance_method")

to

Welcomer.method("an_instance_method").call

Although it doesn't change anything, I think the 5th line of code at the top of the page should be changed from:

Welcomer.method("an_instance_method")

to

Welcomer.method("an_instance_method").call

The following code is incorrect:

red_fish = doc.children[0].children[3].children[1].children[1]

The correct code is the following:

red_fish = doc.children[1].children[3].children[1].children[1]

Note: doc.children[0] refers to the text '
' (i.e., the newline between "%{" and "<acquarium>" specified in the assignment to the variable xml on the top of page 378)

The following code is incorrect:

red_fish = doc.children[0].children[3].children[1].children[1]

The correct code is the following:

red_fish = doc.children[1].children[3].children[1].children[1]

Note: doc.children[0] refers to the text '
' (i.e., the newline between "%{" and "<acquarium>" specified in the assignment to the variable xml on the top of page 378)

doc.children[0] is wrong

doc.children[1] is correct

doc.children[0] returns => "
"
doc.children[1] returns => <acquarium> ... </>

doc.children[0] is wrong

doc.children[1] is correct

doc.children[0] returns => "
"
doc.children[1] returns => <acquarium> ... </>

The 2nd URL seems to be:

http://wiki.rubyonrails.org/rails/pages/ActiveRecord

instead of:

http://wiki.rubyonrails.org/show/ActiveRecord

9460) Heading reads Discussion, but probably should be Solution.

Invocation "./script/generate controller status index" should be
"ruby script/generate controller status index"

Similar error on p.559 where "./script/server" should be
"ruby script/server"

Invocation "./script/generate controller status index" should be
"ruby script/generate controller status index"

Similar error on p.559 where "./script/server" should be
"ruby script/server"

In the last code of recipe 15.7, "require 'active_support/core_ext'" should be replaced by "require 'active_support'".

If "core_ext" is directly loaded, the search path will not be correctly set and "xml_simple" will not be found. Actually "active_support" will automatically load "active_support/core_ext".

So the correct requires are:

require 'rubygems'
require 'active_support'

In the last code of recipe 15.7, "require 'active_support/core_ext'" should be replaced by "require 'active_support'".

If "core_ext" is directly loaded, the search path will not be correctly set and "xml_simple" will not be found. Actually "active_support" will automatically load "active_support/core_ext".

So the correct requires are:

require 'rubygems'
require 'active_support'

form_tag and form_for always go inside "<% %>" tags AFAIK. They used to go
into "<%= %>" tags, tho I don't remember when this change took place (somewhere
between AWDR v. 1 and v.2)

form_tag and form_for always go inside "<% %>" tags AFAIK. They used to go
into "<%= %>" tags, tho I don't remember when this change took place (somewhere
between AWDR v. 1 and v.2)

During the third run of the divide.rb script, the error output says the script
that errored was 'divide_buggy.rb' in three spots of the exception.

In effect, the calling script and erroring script should be the same. Like such:

$ ruby -d divide.rb
Dividing 53 by 0
Exception `ZeroDivisionError' at divide.rb:6 - divided by 0
divide.rb:6:in `/': divided by 0 (ZeroDivisionError)
from divide.rb:6

In order to get the results described by --debug switch, the user must pass the
switch to Ruby and not the script.

All instances that look like '$ ./divide.rb --debug' should be
'$ ruby --debug divide.rb' otherwise the interpreter never enters the debug state.

During the third run of the divide.rb script, the error output says the script
that errored was 'divide_buggy.rb' in three spots of the exception.

In effect, the calling script and erroring script should be the same. Like such:

$ ruby -d divide.rb
Dividing 53 by 0
Exception `ZeroDivisionError' at divide.rb:6 - divided by 0
divide.rb:6:in `/': divided by 0 (ZeroDivisionError)
from divide.rb:6

In order to get the results described by --debug switch, the user must pass the
switch to Ruby and not the script.

All instances that look like '$ ./divide.rb --debug' should be
'$ ruby --debug divide.rb' otherwise the interpreter never enters the debug state.

The code given completes straightaway, not after 5 seconds as described in the
text and simulated output.

Either change the described output to fit the code, or change the code to fit the
output -- perhaps like this:

start_time = Time.now

The code given completes straightaway, not after 5 seconds as described in the
text and simulated output.

Either change the described output to fit the code, or change the code to fit the
output -- perhaps like this:

start_time = Time.now

The code runs as presented, but if you run the block over 5 times - like this:

1.upto(8) do |i| # <-- changed from 5
pool.dispatch(i) do |i|
print "Job #{i} started.
"
sleep(5-i) # <-- the bug
print "Job #{i} complete.
"
end
end

You get something like:

johnperkins:compare_files jperkins$ ruby thread_test.rb
Job 1 started.
Job 2 started.
Job 3 started.
Pool is full; waiting to run 4...
Pool is full; waiting to run 5...
Pool is full; waiting to run 6...
Pool is full; waiting to run 7...
Pool is full; waiting to run 8...
Job 3 complete.
Job 4 started.
Job 2 complete.
Job 5 started.
Job 5 complete.
Job 4 complete.
Job 7 started.
Exception `ArgumentError' at thread_test.rb:54 - time interval must be positive
Exception in thread #<ThreadPool:0x262b4>: time interval must be positive
Job 6 started.
Exception `ArgumentError' at thread_test.rb:54 - time interval must be positive
Exception in thread #<ThreadPool:0x262b4>: time interval must be positive
Job 8 started.
Exception `ArgumentError' at thread_test.rb:54 - time interval must be positive
Exception in thread #<ThreadPool:0x262b4>: time interval must be positive
Job 1 complete.
johnperkins:compare_files jperkins$

Apparently time travel is an ArgumentError.

The code runs as presented, but if you run the block over 5 times - like this:

1.upto(8) do |i| # <-- changed from 5
pool.dispatch(i) do |i|
print "Job #{i} started.\n"
sleep(5-i) # <-- the bug
print "Job #{i} complete.\n"
end
end

You get something like:

johnperkins:compare_files jperkins$ ruby thread_test.rb
Job 1 started.
Job 2 started.
Job 3 started.
Pool is full; waiting to run 4...
Pool is full; waiting to run 5...
Pool is full; waiting to run 6...
Pool is full; waiting to run 7...
Pool is full; waiting to run 8...
Job 3 complete.
Job 4 started.
Job 2 complete.
Job 5 started.
Job 5 complete.
Job 4 complete.
Job 7 started.
Exception `ArgumentError' at thread_test.rb:54 - time interval must be positive
Exception in thread #<ThreadPool:0x262b4>: time interval must be positive
Job 6 started.
Exception `ArgumentError' at thread_test.rb:54 - time interval must be positive
Exception in thread #<ThreadPool:0x262b4>: time interval must be positive
Job 8 started.
Exception `ArgumentError' at thread_test.rb:54 - time interval must be positive
Exception in thread #<ThreadPool:0x262b4>: time interval must be positive
Job 1 complete.
johnperkins:compare_files jperkins$

Apparently time travel is an ArgumentError.