Right. C is a power tool. You can use it to create something blindingly fast. You can also use it to create something dog slow. I've seen Smalltalk programs outperform C programs. I know of an instance where a coworker benchmarked a C implementation of a block cipher vs. a Smalltalk implementation, and the Smalltalk ran 3% faster. Why? Naive manual memory management in the C implementation.
Another classic HLL vs. C argument: the anecdotal slow C program. Clearly, of these, there are many great examples. Unfortunately, it's a crappy argument, because there are also a zillion incredibly slow HLL programs.
I don't think stcredzero meant that as an argument that C is worse than HLL -- it was just a warning that, as with any tool, it can be improperly used.
I'm certainly not against C. I've just used C to write an embedded controller for doing real-time mixing of PWM control signals and laser gyroscope data for a model tilt-wing aircraft. (Which should be able to convert between horizontal and vertical modes in-flight.) I wouldn't use Smalltalk for that. (Yet. If I had 10X more CPU to burn, maybe.) Likewise, I'm not going to write my next Commodities Trading app in C.
That is a false dilemma tho'. No-one in the Python world has anything invested in a 100% Python solution to anything. Doing the compute-intensive bits in C is in fact expected and encouraged! The same is true in the Tcl camp. Maybe some HLL communities (Java?) like to be "pure" but I've not ever encountered that.
Java communities typically favor portability (and easy deployment) over what you can gain in terms of speed. A lot of libraries/frameworks actually even advertise themselves as "100% pure java", whereas one might expect "expensive bits optimized in C" as attractive as well (but you rarely see it).
I'd like to see the c code to support this, it's very hard to imagine anything you could do this wrong in something like a block cipher that would ever make it as slow as smalltalk. I'm sorry, I just don't buy this...
The C DLL was one of RSA Data Securities reference implementations from about a decade back. The Smalltalk guy could ask the VM implementor for goodies -- like 32 bit and 64 bit bit-arrays and various primitive operations on them.
So long as you keep an entire algorithm in one method, and restrict yourself to certain optimized operations, the resulting JIT-ed code will look like it was produced by an unoptimized C compiler. There will be no Smalltalk message sends. (This approach is a bit fragile, though, since you have to have esoteric knowledge to maintain code like that and keep the same level of optimization. Most of the time, you'd just implement a speed-critical operation like this as a DLL in C to be called from Smalltalk. Smalltalk/X has a different approach -- you just inline C code in your Smalltalk method.)
Combine that with naive memory management in the C code, and you have C code that loses a benchmark. Mind you, that was bad C code -- reference code only has to be correct, not fast.
"Smalltalk is Slow" usually a sign of ignorance or trolling. For lots of problem domains, the advanced commercial Smalltalk VMs are pretty darn fast.
I'm not saying Smalltalk is slow, but I've worked with rsaref since the mid-90s, and I don't remember a block cipher routine in it that was horribly slow due to allocation.
So instead of Highly Optimized Smalltalk (with VM implementor help) = Horrible C, we have Highly Optimized Smalltalk (with VM implementor help) = Average C.
