I really want to learn C, like he says. I get plenty done without knowing it, and I have few doubts I can continue to find decent work without knowing it, but I haven't been able to gain any traction when I try to learn it.
I've got the books sitting front of me, and I've written some trivial visualizations of sorting algorithms using terminal output, but damn if I can find a way to use C as a web developer. If there were just some use case where C would help me get something done, I'd be all over it.
For years as I developed with Python and Javascript I thought "I really should write something in C again and maybe I should learn some C++". But nothing I was doing was performance critical or low level enough that I needed C or C++.
Then, a funny thing happened and I started playing with the Arduino.
And, while they go to great pains to hide it from non-technical people, the whole Arduino stack (ignoring the IDE) is based on C/C++. So then I began writing C again and learning C++. (I always find it slightly odd that I'm using C++ on a microcontroller--admittedly it's very small subset of C++, mostly just used for the object encapsulation/abstraction. (Insert handwavy, "yes, yes, I know it's not strictly C above library level" here.))
So, you might like to take a look at playing with the Arduino if you want to get into C. And if nothing else it's fun playing with tangible things that interact with the "real world".
You can take slightly larger leap, and program the Arduino processor directly, without using the Arduino abstraction. It's the next logical leap, as "industry" doesn't really use the Arduino platform.
A great exercise for Web developers is to build a minimal web-server. You can choose your definition of 'minimal' but essentially it should receive some request and return a response (from and to a web browser). As a webdev I found even the very, very minimal web server in Land of Lisp to be really insightful to build. If you wanted to flesh out the actual handling of http verbs better that would be even more awesome. Minimal webservers should be the minimal lisps of web developers
> I've got the books sitting front of me, and I've written some trivial visualizations of sorting algorithms using terminal output, but damn if I can find a way to use C as a web developer
Write a C-extension for infrastructure-level ruby gem if you're a rubyist, or fix a bug / add a feature in your favorite scripting language.
So, if you're a web developer, you probably use Perl, Python, Ruby, or some similar language. Guess what, these are written in C(+). When you need to add or fix a feature, guess what language you use?
(+) Yes, yes, programming languages are not written in anything. Their implementations are. I know you know what I mean.
I hear you, but adding or fixing a feature in Python/Ruby/Perl doesn't ever come up for me. Does it for you? Plus, I understand that much of the standard libraries are written in the high level language.
See if you can find some bottleneck in one of your Python/Ruby/Perl apps that's chewing a lot of CPU time. See if you can devise a definition for a relatively simple function that if it was REALLY FAST would make this hot spot a lot more efficient. Then implement that function in C and expose it to the high level language.
It might not be a good idea to actually use your fast C function in production (if you're new at C you might have a security vulnerability in there, and there could be maintenance headaches). But you'll have learned a lot.
> See if you can find some bottleneck in one of your Python/Ruby/Perl apps that's chewing a lot of CPU time.
I found that hard to do as a starting developer who didn't know what was actually a bottleneck in my code, but I did have a particular library/gem that I wrote myself that did a few things I constantly needed (think of authentication in a web app, it's almost always included). So I rewrite it in C. It didn't really make it faster, or better, but I learned a bunch of C.
I have not needed to add or fix a feature in a programming language but I have needed a ruby gem that uses native extensions (written in C) that are not available for my environment. I know enough C, but I have no idea how to implement this so I can't use it. Thus, instead of calling a few functions in a library I literally have to install another version of ruby, wrap the gem in some high level API, and expose the service via Sinatra, and call it via Net::HTTP. Not fun.
Particularly, libraries search their catalogs using the z39.50 protocol. Ruby 1.8 has a gem ruby-zoom[0] that does this using YAZ[1] however there is not a ruby 1.9 version. If anyone has a guide that explains native extensions for Ruby, please advise.
Try writing a custom apache module? (Or nginx, or similar...) You can do some pretty fun things once you get into that layer of the web transaction life cycle (e.g. modifying behavior of things outside of the usual request/response part of the http transaction that you'd usually work with).
I'm learning it now too. My primary use case, at least initially, is going to try to extend python with C. Not really because I have a need, but mostly because I can't really imagine needing C for any other task than speeding up some of my poorly optimized python at some point down the road.
Listening to you, I'm thinking about just reading the C books and nothing more. If C's usefulness to non-C-programmers is simply that it gives them greater understanding of how computers work, then reading might suffice.
I'd encourage you to write some code, too. There's a lot of intuition you gain by manipulating the bits that you can't get by reading. You may think you know what a pointer is, but until you've had to debug a double-free segfault or a memory leak, it's hard to understand the ramifications of the concept. The standard string handling functions look trivial in documentation, but when you actually use them you'll find yourself forgetting to add null bytes or not realizing that some of the functions mutate the string you're working on. I'd argue it's that process of figuring out how to make things work that gives you the famed understanding of how computers work.
I don't think sufficient understanding of how computers work can come from just reading a book. If it could, you should read a book about how computers work, instead of trying to glean that through the lens of C programming. Understanding comes from doing.
Personally, I don't agree. I think a majority of the value of K&R actually comes from the exercises.
K&R gives you a very good, clear and detailed explanation of the topic at hand and then immediately gives you several exercises to practice it. If you don't do the exercises the information will just fall out of your short-term memory whereas if you actually do them your long-term memory will get engaged as well as general muscle-memory for C coding.
C is pretty flexible. Just write a create_pool() function, an sprintf()-workalike that allocates memory from the current pool, and a destroy_pool() function that deallocates the current pool. Use one pool per request. It's not Perl, but I'd imagine it's not that inconvenient - and very fast.
(See also: raphters, which was featured on HN not so long ago.)
I've got the books sitting front of me, and I've written some trivial visualizations of sorting algorithms using terminal output, but damn if I can find a way to use C as a web developer. If there were just some use case where C would help me get something done, I'd be all over it.