Programming in Python 3 (First Edition)

Issue	Resolution	Reporter
Page 21. Clarification.	The book shows the following interactive session: `>>> a = "many paths" >>> b = "many paths" >>> a is b False` In some cases, comparing the identity of two strings or numbers, even if assigned separately as shown here, will return `True`. This is because some implementations of Python will reuse the same object (since the value is the same and is immutable), for the sake of efficiency. Of course, this is a reminder to compare values using `==` and `!=`, and to only use `is` and `is not` when comparing with `None` or when you really do want to see if two object references, rather than their values, are the same.	Robert Lummis; Leon Matthews
Page 54. Python 3.1 difference.	The floating-point number output shown on this page is different between Python 3.0 and 3.1. This is because 3.1 uses a smarter algorithm for displaying floating-point numbers that shows as few digits as possible without loss of accuracy. Note though, that internally, Python—like practically every other language—stores most floating-point numbers as approximations, so the tiny amount of inaccuracy is still there, just less obvious.	Alexis Layton
Page 56. Incorrect word.	In Table 2.5's entry for `math.floor()`, replace "smallest" with "largest".	Benjamin Fischer
Page 57. Erroneous table entry.	In Table 2.6, delete the `math.sum()` entry, and in the index delete the `math.sum()` and `sum() (math)` entries. (The correct function is `math.fsum()` which is listed in Table 2.5, and is in the index.)	Ihar Hrachyshka
Page 66. Slightly misleading explanation.	The third paragraph describes the two colon slicing plus striding syntax, and says that you cannot omit the step—it should have said that you can omit it, but there's no point doing so. This is because if you omit the step, it defaults to 1, but that's the default anyway, so for a step size of 1, the one colon syntax is sufficient.	Tom Zych
Page 71. Incorrect code.	The code snippet at the bottom right of the page should have the line `result = "", "", s` not `result = s, "", ""`.	Takahiro Nagao
Pages 82 & 88. Error in footnotes on both pages regarding Terminal.app on Mac.	The footnote on each of these pages says that Mac uses the Apple Roman encoding in Terminal.app and therefore cannot output all the Unicode characters. This is incorrect—happily the Mac uses UTF-8 and both the `print_unicode.py` and `quadratic.py` examples work fine on the Mac. (I made the mistake because I tested them on the Mac by running the book's test suite—which runs tests using the `subprocess` module—rather than directly at the console, and `subprocess` seems to use the Apple Roman encoding on Mac.)	Author
Page 94. Converting the text to HTML and truncating it has a subtle bug.	The code at the top of this page truncates HTML-escaped text, which means it might erroneously cut off the end of an HTML entity. Replace the top 6 lines with these 5 lines: `if len(field) <= maxwidth: field = escape_html(field) else: field = "{0} ...".format(escape_html(field[:maxwidth])) print("<td>{0}</td>".format(field))` Note also, that the single line of code quoted in the middle of the page suffers from the same problem and should be replaced with: `print("<td>{0}</td>".format(escape_html(field[:maxwidth])))`. (If you are using Python 3.1 or later you can use `{}` instead of `{0}` throughout.)	Peter Bray
Page 117. Error regarding comparison operators.	The last paragraph says that dictionaries can be compared with `<`, `<=`, `==`, `!=`, `>=`, and `>`. This was true in Python 2, but is (fortunately) not the case in Python 3 where only `==` and `!=` are supported.	Jennifer Lindner
Page 121. Incorrect word.	In the second paragraph's second sentence, beginning "And we can use the union operator", replace "union" with "intersection".	Benjamin Fischer
Page 136. Incorrect number.	In the paragraph immediately above the Copying Collections subsubsection, the returned list should start with `-7.5` not `-7.3`.	Benjamin Fischer
Page 144. The `calculate_mode()` function works correctly but is needlessly complicated.	The `calculate_mode()` function only ever deals with integers, so it can be simplified. Change the list comprehension's `if` condition from `math.fabs(frequency - highest_frequency) <= sys.float_info.epsilon` to `frequency == highest_frequency`. This change also affects the text, so replace the second sentence in the second paragraph that follows the code snippet (the one starting "Since the numbers may be floating-point"), with "We can compare using operator `==` since all the frequencies are integers."	Tim Pietzcker
Page 152. Slightly misleading description of using `return` in a `for` loop.	Replace the last sentence of the penultimate paragraph—the paragraph begins "If a `continue`", and the sentence to replace begins "If the loop is broken"—with: "If the loop is broken out of due to a `break` statement, or a `return` statement (if the loop is in a function or method), or if an exception is raised, the `else` clause's suite is not executed. (If an exception occurs, Python skips the `else` clause and looks for a suitable exception handler—this is covered in the next section.)"	Peter Bray
Page 157. Incorrect exception mentioned.	In the second paragraph the exception mentioned should be `UnicodeDecodeError`, not `ValueError`	Takahiro Nagao
Page 168. Wrong units of measurement.	For the `heron2()` function the units of measurement should be "square meters" and "square inches" (not plain meters and inches).	Peter Bray
Page 205. Incorrect word.	In the first paragraph change `--width` to `--maxwidth` (twice).	Takahiro Nagao
Page 231. Subtle incompleteness.	The third paragraph from the bottom says that custom classes support `==` by default and that the comparison always returns `False`. There is in fact one case where the comparison returns `True`—when a custom object is compared with itself.	Alexis Layton
Page 296. Incorrect class name.	In the first and second paragraphs, change `textwrap.TextWrap` to `textwrap.TextWrapper`.	Takahiro Nagao
Pages 398 & 399. Localization issue.	The code at the bottom of page 398 and the first line of code on page 399 may not work in all locales. This is because the `subprocess` module always uses binary data but encodes/decodes using the local encoding—this has been reported as bug 6135 and should be fixed in due course. If you get burnt by this the solution is to set `sys.stdin` to binary mode and do the decoding yourself. For all Python 3.x versions this can be done by reading the data with `data = sys.stdin.buffer.read()`; and for Python 3.1 and later it can be done by `sys.stdin = sys.stdin.detach()`, and then doing `data = sys.stdin.read()`—and in either case, decoding `data`, e.g., `lines = data.decode("utf8").splitlines()`.	Takahiro Nagao
Page 409. Incorrect URL.	In the fourth paragraph, change the URL to code.google.com/p/python-safethread.	Takahiro Nagao
Pages 421 & 422; an explicit context manager is not needed for `gzip` with Python 3.1.	From Python 3.1, `gzip.GzipFile` supports the context manager protocol so it is not necessary to use `contextlib.closing()` or a custom context manager when handling a `gzip` file handle with Python 3.1—unless you want backward compatibility with Python 3.0. The techniques shown are still valid though, e.g., for classes that don't support the context manager protocol, and of course for programs that must support `gzip` for all Python 3.x versions.	Author
Page 434. Subtle bug.	The `find_dvd()` function has a subtle bug. The fix is to replace the `enumerate()` line with these lines: `matches = sorted(matches, key=str.lower) for i, match in enumerate(matches):`	Takahiro Nagao
Pages 461 & 462; typos.	Change every occurrence of "W." to "W" (e.g., in the name "James W Loewen")—or, alternatively, change the regex on page 459 to: `name = re.sub(r"(?P<forenames>\w+\.?(?:\s+\w+\.?)*)\s+(?P<surname>\w+)", r"\g<surname>, \g<forenames>", name)`	Peter Bray

First Edition: Errata for the First Printing only (additional to the errata above—these have been fixed in the Second and subsequent printings)

Issue	Resolution	Reporter
Page 4. Package names and availability are not as stated.	Only two installers appear to be available for Windows, a generic one (x86) and an AMD64-specific one; use the generic one unless you are sure you have an AMD64 processor. Fedora 10 does not have Python 3 packages after all, so you must manually build from the sources. Ubuntu 8 does have Python 3 packages—their names are `python3.0` for Python 3, `idle-python3.0` for IDLE, and `python3.0-doc` for the documentation.	Author
Page 5. Executable names are not as stated.	The Python executable is called `python3.0` (or `python3.1` depending on which version you installed) and IDLE is called `idle`. In the book the names used throughout are `python3` and `idle3` to stand for whatever the actual names are (and of course you could alias or soft link to make these names work—in fact, some packages do this for you).	Author

Programming in Python 3 (First Edition)

Errata

ISBN-10: 0137129297 – ISBN-13: 978-0137129294