09 March 2011
A friend of mine pinged for some good bootstrap content on various technologies, namely Perl, MySQL, PHP, Python, Linux, C and C++. I learned (or, tried to... I'm convinced I'll never understand C++ and Perl) all of these different things over the years and so my knowledge is either outdated or incomplete. Nevertheless, I figured some of it might help others, too, so I'm reprinting a cleaned up version of my response email here. This information, of course, reflects my biases and experience and, except Perl, reflects professional experience in some form or another.
Perl: Never. ever. use. Perl. I don't know it (couldn't seem to sit and learn it with a straight face. It made me anxious), and can't recommend it unless you're a hard core system administrator who doesn't know Python, Ruby, C or Bash. It's horridly unreadable to all but the most exacting of eyes. It's been stuck at version 5.x for the last 10+ years. 6.0 has been "in progress" since 2001, code-named "Parrot." When Parrot becomes real I could easily see taking it on again.
C: C is markedly more difficult to learn than PHP and at the same time much friendlier for the simple things. (Brian) Kernighan & (Dennis) Richie wrote the introduction to C, back in the 70's. People still refer to the variant of C as "K&R," or they refer to a particular standard. C99, for example, is the standardized version of C that came out in 1999. I only know "C99"-ish C, not proper "K&R" C. I only write C when I'm trying to glue stuff - typically operating-system specific stuff - together: Java and Windows, OSX, or Linux, via C, for example, or Python and something else, via C. I rarely have a use case for a full, stand-alone C program, if I'm honest. I almost always write modules that use C. C has no library - just a few standard types and the very, very loose language itself. So, the language itself is very easy, if a bit useless. You'll need to bind to something - either to your operating system (i.e., through POSIX or some additional APIs, or to some other library) to do anything. I've only met one person I'd call a "master" at C. To learn C, I learned by Googling and spending lots of time looking at Linux sources (sources included with Linux, not only the source for the Linux kernel itself) and so on. Check out SourceForge or Github.com.
C++: I tweeted recently that Herbert Schildt's now outdated C++ book was available for free download, which you might try finding. I think it was on MSDN (the Microsoft Developer Network). I learned C++ from Bruce Eckel's amazing "Thinking in C++" back in the 90s. He also wrote "Thinking in Java," which I loved so much so that I wrote in to him and gave a positive reader comment back in 2001. My comment is on one of the few pages on the inside front cover of his book, book under "reader comments" to this day ;-) He used to make the books available for free download, which I suspect you could still find if you Googled. Otherwise, you can buy the two volumes of it on Amazon. In college I also read "Starting Out With C++" which was... lame compared to "Thinking in C++," but it's also thorough (and shorter!), so it might be a good intro-read to ramp you into it. As with C, but to a lesser extent, to make use of C++, you need third party libraries since C++'s core standard library is kinda crap compared to Java or .NETs or Pythons, etc, so one popular choice is to use a platform abstraction like Boost, or Apache Portable Runtime (APR) or Netscape Portable Runtime (NPR). Some good books on Boost/APR that I often refer to these days when I've (rare) occasion to write C++ is "Beyond the C++ Standard Library" and "Cross Platform Development in C++." The first introduces Boost, the second introduces ... APR and a lot of other stuff. All that said, C++ is not as universally used (or even close to as universally useful) as C, and it's far, far, far, far more complicated than Java or C. I would avoid C++ if possible. The only reason I liked C++ is because I then understood why Java was an improvement.
Linux: I have no good book on Linux to recommend, only several that all helped fill in parts of the puzzle: Red Hat Linux Administrators' Handbook, O'Reilly's UNIX Power Tools, Practical MythTV, EXTREMETECH "Hacking Ubuntu", EXTREMETECH "Linux Toys" I and II, the "Official Gnome 2 Developer's Guide," etc. Of course, with this one I was lucky to have a talented Linux ninja - Mario Gray) - to really push me into it and make sure I didn't have a chance to escape until I'd seen everything he wanted me to see. By then, I didn't want to escape. Linux worked wonderfully. ;-)
MySQL: MySQL's dead simple.
sudo apt-get install mysql-server-5.1 on Ubuntu Linux, and then follow any tutorial on the internet. It's a a database with 3/4 of the features of PostgreSQL and Oracle. It's also a ...scary open source project at the moment, because of all the politics surrounding licensing. MariaDB, MySQL, Drizzle, Percona, etc. I know Google has patched it over the years to suck less. If you need scalability and speed, use a NoSQL option. If you need features and an amazing, open-source database, use PostgreSQL. If you need a little bit of both AND commercial support, use Oracle, or the EnterpriseDB PostgreSQL build. Don't use MySQL unless you can avoid maintaining it. Amazon and Google App Engine for Business provide MySQL-like databases for their clouds, for example.
Python: Python was one of the easiest languages to learn since it was designed to be easy to learn, and, if I recall, Guido van Rossum (Python's creator) had experience with ABC, a learning language, which influenced Python's creation. I got my first look at Pyhon in the late 90's, or early 2000's. I think Bruce Eckel, again, mentioned Python in one of his books and that was enough to catalyze me to learn it properly. It's not hard to get going quickly, and for many years I kept it as my in-my-backpocket language, to do real work when Java might've been too complicated or if the task leant itself particularly well to Python's strengths in leveraging operating system-specific tools and APIs. Python has a large part of the POSIX APIs almost directly exposed through Pythonic APIs, for example, so your C skills come in handy there. It's a very good glue. I haven't really done more than played with Python's more recent web-tier support (and WSGI API, which provides the gateway interface between a web framework and application to the underlying web server, in much the same way servlets sits on top of the webserver, but below Java-based web applications.). I did have a pretty good experience with Zope (and the Plone CMS), which sort of predates all the current crop of Python web-framework killer-apps like Django and Turbogears. I have an older version of O'Reilly's "Programming Python," and I saw that it's since been updated, but it's huge - and as much a tutorial as a reference.
So, that's a lot of how I got started. I'd of course be interested in any suggestions since a lot of this is outdated (At least, I hope it's outdated and that better texts have come since!).