Sponsored Link •
A couple of weeks ago I was asked to review in my blog "Expert Python Programming" by Tarek Ziade' in exchange for a free copy of the book. I accepted, I received the book a few days ago and I read it in the weekend. Here are my impressions.
Expert Python Programming - as the title implies - is intended for proficient Python programmers who want to become experts. In my opinion the target audience is very well chosen, since we do not need yet another beginners' book, whereas there are very few Python books intended specifically for skilled programmers.
The book fullfills its goals exceptionally well: it definitely gives you enough hints and suggestions to turn you from a mere practictioner into a professional Python programmer (of course some effort on your part is required in order to reach that goal, but a book cannot perform miracles ;).
In only 350 pages Tarek Ziadé covers an extremely wide range of topics. Expert Python Programming is not just a book about Python, the language; it is also a book about Python, the platform: it covers all the major libraries and tools used to write and to deliver profession Python application. It is the knowledge of such tools and techniques that make the difference between an amateur and a professional developer.
Moreover, the book is a collection of best practices for modern software development, covering topics such as how to use a Distributed Version Control System, how to setup a buildbot for your tests, how to setup an automatic build of your documentation, etc. Really a lot of good stuff that I have never found collected together in a single place.
The thing that impressed me the most is how much this book is updated: basically it covers all the latest development in the Python community, even very recent ones like the Sphynx documentation system, to name just one which I think is extremely interesting and little known.
I like very much the overall attitude of the book, especially the lack of verbosity. There is a lot of really good information in a surprising small number of pages, since as the author says, "this book is not intended to replace inline resources, but rather aims at complementing them". Other books I have seen contain thousands of pages of redundant information you can find in the manuals, whereas Expert Python Programming is more like a guide through the jungle of the Python resources, just what a competent programmer needs.
I follow the latest developments in the Python world very closely, so I did not expect to find something I did not know already, but I was very pleasantly surprised to discover I was wrong.
Basically the book focuses on three topics: how to use Python at its best (about new and/or advanced Python constructs), how to write a professional Python application (with tips about IDEs, DVCS, testing frameworks, optimization techniques, etc) anf how to deliver and mantain an application (how to document the code with reStructuredText and Sphynx, how to distribute it with setuptools and other tools such as PasteScript, zc.buildout, builbots, how to set up a bug tracker with Trac, and more).
The first chapter, about installing Python and setting up a confortable development environment, risked to be removed: as the author says in the preface "I almost removed this chapter since the book is not intended for beginners". I am glad that it was was kept at the end: there is useful information there even for an expert Pythonista like myself (let me be a little immodest here).
For instance, there are tips about how to get a nice working environment under Windows by using MinGW and MSYS which I have never used (I do not develop under Windows; in the past I have used Cygwin which is a little bit too pervasive IMO). If I needed to develop something on Windows in the future I would certainly take advantages of Tarek's tips.
I liked the chapter about setuptools, since I learned a lot about how to make eggs for distributing Python code (at work all of our code is for internal usage, we do not distribute it, so I was pretty much ignorant about the subject). I plan to take advantages of setuptools to distribute a couple of Open Source libraries I am working on. I must confess that the setup.py script for my own decorator module was contributed by an user, since I did not know how to use setuptools at the time.
There is an entire chapter about choosing good names, and I am glad there is, since choosing good names is possibly the most difficult thing in programming. Moreover, the chapter gives a very good introduction to Python code style guide, the (in)famous PEP 8, and it cites Pylint, a tool every Python developer should know about (the other is pychecker, which is not cited, so I will cite it here ;).
Whereas the chapters about coding conventions and documentation are for everyone, the book contains good advice for experts too: they will enjoy the chapters about profiling and debugging. There is even a paragraph about Guppy and Heapy, a tool to track memory usage I was never able to use (as the author says "Heapy is not easy to use and needs a bit of practice"). For people working with databases, there is also a nice introduction to SLQAlchemy, the major database toolkit for Python.
The last chapter about patterns is definitely well written and it may appeal to pattern aficionados. I personally I am an not estimator of patterns. Instead of motivating my position, I will just report a couple of citations from comp.lang.lisp (you may substitute Lisp with Python):
I remember when I read the original "Design Patterns" book, thinking "wow, non-Lisp languages sure make you jump through all sorts of hoops to do things we take for granted." That's why they need to give names to all the patterns, because you can't "just do it." -- Barry Margolin
Yeah, I remember having the same reaction - kind of sad when a language is so broken that there is a list as long as your arm of collected workarounds that are glorified as "design patterns." -- Raffael Cavallaro
I must confess that I liked a lot the chapter about the MRO, since my own paper is extensively cited there as "the reference document" about the subject. I certainly would not have expected such a result when I wrote it six years ago. At the time, I was just a hobbyist programmer who had freshly learned OOP and I wanted to give a very minor contribution to the Python documentation, since I could not give a contribution in code. It turns out, however, that there are very few programmers who are willing to give contribution to the documentation, so that whatever you write becomes "the reference document". As a consequence, nowadays, every Python programmer who wants to call himself an expert (and there thousands of them) has to read my paper: frankly, this is quite flattering ;)
Moreover, Tarek also talks about metaclasses and he cites the first paper of our (our here means David Mertz and me) trilogy. However, I would have cited the original version appeared in IBM DeveloperWorks and not the republished version appeared on OnLamp.
I am glad about those citations since it looks like the time we spent in documenting such arcane features was well spent, somebody actually read the result of our efforts, and we did actually contributed something back to the community.
On the other hand, Tarek writes a lot (and very well) about decorators but he does not cite my own own decorator module. Helas, you cannot win always ;) It does a very good job with descriptors instead.
Tarek also gives very sound advice about tricky Python constructs, such as Python super (use it consistently) and multiple inheritance (do not use it). I also liked his advices about testing (both unittest and doctests) and documenting. Tarek is also a great fan of Documentation Driven Development, an idea I started advertising at least from 2003.
As a final note, I would like to see a couple of words about the publisher, PACKT. I must admit I did not know it, probably because they entered in business only few years ago and they do not sell many Python books. However, I have looked at their web site now, and I see that they have a really large offer especially about Web Development books. If all their books are of the same quality as "Expert Python Programming", it is certainly a worthy publisher. I have not seen a single mispell in the entire book; moreover the paper, the cover, the pictures, the fonts and in general the look and feel are just fine.
The book is certainly worth its money and more. It is also good to know that part of the revenue will go to the Python Software Foundation. In general I like very much their support for Open Source and their marketing strategy: it makes a lot of sense to me, I sincerely hope their business model will pay off, as it seems to be the case.
All in all, Expert Python Programming is the best book in its category I have ever read. I definitely recommend it and I will ask the company to buy one for the office library. You are adviced to do the same!
|Michele Simionato started his career as a Theoretical Physicist, working in Italy, France and the U.S. He turned to programming in 2003; since then he has been working professionally as a Python developer and now he lives in Milan, Italy. Michele is well known in the Python community for his posts in the newsgroup(s), his articles and his Open Source libraries and recipes. His interests include object oriented programming, functional programming, and in general programming metodologies that enable us to manage the complexity of modern software developement.|