On Dec 28, 2007 7:16 AM, Gary Baluha <[EMAIL PROTECTED]> wrote: > On Dec 28, 2007 7:51 AM, Erik Wikstrvm <[EMAIL PROTECTED]> wrote: > > > On 2007-12-28 07:33, Brian Hansen wrote: > > > > [snip] > > > > > Is he right? > > > > Yes and no. First of all you should realise that Linus and most other > > other kernel hackers are biased. When it comes down to it C++ is not a > > better or worse language to write a system in than C, it is just a > > question about how you use it. > > > > But as Linus wrote, there are a number of people out there who learned > > to program in Java and then learned C++ and now think that they can do > > some low-level programming. Truth is they can not since they have > > learned to think about programming in a very high-level way. Someone > > once said something like "nothing is impossible with enough layers of > > indirection", but none ever claimed that many layers of indirection was > > efficient. If you are down in the kernel writing code that will be run > > while holding a lock every micro-second counts since you are stalling > > the progress of other threads/processes, so you want efficiency. > > > I think that's one of the main problems with programmers today. All the > schools I know that teach programming, start by teaching Java to the > students. It's definitely easier to learn Java with little-to-no prior > programming experience, than it is to learn C (I struggled for quite some > time to fully understand and realize how C uses pointers, in all its various > ways). I think the problem comes when these same programmers see that Java > and C have a similar style of syntax, and think that simply learning the > superficial differences means they now understand the language. Two > programming languages can look similar on the surface, but be very different > beneath the skin.
There is a community college in the San Francisco Bay Area area that encourages learning C before learning any other language. I've taken intro to C and intermediate C. I didn't do too well with pointers either. It's tough working full time and attending class twice a week and studying and living. Heh. > > I'm also not sold on the concept of object oriented programming in general. > Along the lines of "nothing is impossible with enough layers of > indirection", I think too much abstraction also removes the programmer from > what he/she is actually doing. OOP is a nice concept on paper, until you > realize that most large programming projects are not written by a single > person, and most people will not be coding the entire tree of abstraction > they are using. This leads to errors that may be very difficult (if not > impossible) to fully track down. > > > > > But if you are writing a complicated GUI for an application that will > > run on a 3GHz machine with 2GB RAM a few layers of indirection can be > > very nice, and none will notice the few milliseconds you lose. > > > > It is about choosing the right tool for the job, but also about choosing > > how to use the tool, just because a hammer is the right tool does not > > mean that hammering away is the right way to do things. > > > Yes, that's true, but sometimes using that old trusty tool (C) can be just > as effective as the shiny new tool (Java, C++, etc).
