Thread: How can I learn?

Too much C++ bashing going on;IMHO ........so presenting some defense :
>"f you talk to the C++ crowd they will tell you that C++ is C on steroids."
NO, C++ is a different language, with some similarities to C.

> "to fully deploy C++ code you have to *think differently* than you do with C"
Er.. that is the point of it being a different language....If it does not have a new way to think about problems, it isn't worth learning.

> "C++ strings require conversion before use with OS APIs" :
That would complicate everything..because character arrays and streams are an universal interface...and would be understood by everything. And the conversion is trivial... The input is automatic(by a constructor, and output is by *.c_str() .

> " . I've tried I don't know how many times to get my head around it and just plain can't. Object Oriented Programming bears little or no reseblence to how my head works and it just won't sink in. "

Maybe, the problem is that you're trying too hard ? ...it isn't really radically different.
I find most (if not all I've seen) C++ programs having their outlines as procedural. Classes and objects play a role within that...making the process more modular.

> "Of course your experience might be different but I would never recommend C++ to a new programmer. "
Mine was.... I found/find the C++ standard library easier to learn/use.(Except possibly ... formatted I/O!)
The remaining/basic concepts are same...as long you're careful, the knowledge of one, for the most part, carries on to the other.

> "....and be as compatible with the underlying OS API as possible (able to directly import headers and libraries)... But as far as I can tell from extensive searching, such a laguage does not yet exist."
I think... such an OS API does not exist because of inconsistent design. Though not an OS API, if you have looked at OpenGL, it fits what I think an API should behave like.

>"I've always held that a programming *language* should be as simple as possible (that is, use a minimum number of keywords to guarantee general applicability) be as safe as possible (runtime error checks, exception handline, etc) "
"The next attempt to solve the problems was "managed code" (.NET and JAVA) returning to the bad old days of interpreters and runtime modules in the hope of improving the quality of code. All they really did was to make things yet again more complex for the programmer."

Agreed

Lets see if the rest of the C++ crowd has anything to say.

(@Those-Who-Must-Not-Be-Named : Sorry for 'derailing' the thread... feel free to split !)

Originally Posted by manasij7479

Instead of always badmouthing Open Source, you should try to look at the benefits too.
If you could look at what exactly was wrong in this case, you could have succeeded.

Oh believe me, I've given it a pretty fair trial over the years, even participated in a few projects.

Given some of the bickering and fence building I've seen, I remain convinced that a tight group, or a single person, will almost always do a better job than having people who's primary credential is ownership of a computer messing about in the code.

Have you ever considered how strong and how well standardized Linux could be if it was developed by a core group of specialist coders? Really... you want a dog's breakfast, try making sense of all the Distros and branches of that OS... waddafreakingmess!

By way of contrast... Look at Pelles C... the whole thing was done by one person (Pelle Orinius) and the result is magnificent. It's consistent, complete, easy to use, standards conforming and extremely well documented. Can't say that about most open source work.

Originally Posted by CommonTater

Have you ever considered how strong and how well standardized Linux could be if it was developed by a core group of specialist coders?

If by that you mean, have a core group of specialist coders develop and package the entire OS, that would well and truly suck. I think the linux world is much stronger for its diversity. That does place certain extra responsibilities on the user, but I also consider that a good thing, because god save us from drooling hordes of idiots -- they should turn back at the door, thanks. I'm an elitist that way.

Conversely, I have to use "centrally managed for the masses" windows and OSX occasionally, and IMO they are absolute lowest common denominator crap -- totally inflexible with a ridiculous interface that is the product of marketers, not programmers. OSX is particularly goofy. Windows is just obtuse and awkward.

Anyway, linux proper (the kernel) is developed by a core group of specialized coders. Go tell them, "Oh I heard this is open source so I am free to get involved, be a goof, and screw things up." I am sure they will be enthralled.

Really... you want a dog's breakfast, try making sense of all the Distros and branches of that OS... waddafreakingmess!

Because you don't get it, and no one is really trying to sell it to you. I agree it is probably very shocking coming from the sterile, monotonic world of windows.

Keep in mind that altho some people have made a business out of linux, linux is not in any way shape or form a business, and this is reflected in the nature of "the product". Very positively, IMO, but the contrast is stark. If all you like to eat is stuff you can buy at the corner store in a foil package, putting a fresh vegetable in your hand might well gross you out.

By way of contrast... Look at Pelles C... the whole thing was done by one person (Pelle Orinius) and the result is magnificent. It's consistent, complete, easy to use, standards conforming and extremely well documented. Can't say that about most open source work.

I can say it about most of the open source work I depend upon. Look at the mozilla project compared to the cluster___ know as "Internet Explorer", etc. I can't really say anything for the documentation of either of them, but certainly Mozilla firefox is significantly more standards compliant and generally less defective.

And GCC is hands down the most popular C compiler in the world, for better or worse. Similarly, the internet runs mostly on apache, an open source project.

Originally Posted by dennis.cpp

Guess I chose the wrong keyword here... I would have loved a little more input concerning my "dilemma", instead I caused a (very interesting though) huge discussion on free software.

That happens. I think it's okay (within reason*), since it is not like you have to sit thru a lecture or conversation -- you can just skip posts.

* but who's, lol

If this is your dilemma:

Originally Posted by dennis.cpp

I started several times trying to learn C++, but I don't get the feeling I am moving on in any way. I never had that feeling with C, where I definitely saw progress. Parts of me really *want* to learn C++, but then other voices inside my head always tell me that I'd be better off sticking with C and working my way through GTK+.

Here's $0.02: I learned C before C++, and I still use C much more. If I had learned C++ first, I imagine I would not use C much, but who knows?

The reason I don't use C++ much is because:

1) it consumes much more resources (there's an example below in the bit about gtk)
2) compared to the other formal OO languages I use, which are dynamically typed interpreted languages (and therefore, lower performance) C++ is much more awkward. The syntax is long-winded, and I find some of the idiosyncrasies (such as iostream) silly; they are often IMO not any form of improvement over the C equivalent. Of course, you have the freedom to choose, but I do think it is better, when in Rome, to do as the Romans do, if you want to make sure everything is going to work out, etc.

I believe the major significance of C++ is that it makes a bit of a compromise, resource usage wise, in order to provide a more foolproof high performance setting. That is probably a great advantage for large scale projects involving lots of developers. Unfortunately, I don't do that kind of work, so many of the features to me seem obtuse.

However, it is still interesting and worth learning. I'd be happy to work with it more given the chance, but I don't have much motive to do so on my own. More than once I've had some little thing I wanted to do and thought, hmm, I'll use C++ this time. However, it really doesn't save me much time, and when I see the difference in terms of memory use, I just cannot stomach it and end up starting again in C.

I've though a lot about using C++ just for the OO -- ie, ditching the STL, which is the major memory pig, except that this then comes down to the difference between:

Code:

myobject.mymethod(somearg);  // C++
mymethod(myobject, somearg);  // vanilla C

No difference at all, considering that I really don't care about public vs. private, etc, because of the nature of what I am doing. I don't have to write for a team (or not much of one) or the general programming public, so it doesn't matter to me how "safe and friendly" my API is.

WRT GTK+, I learned that before I learned C++. So when I learned C++, I started using gtkmm, the C++ wrapper library for gtk. Doing exactly the same thing in gtkmm vs. gtk+ will use 300-500% of the memory, because all those handy wrapper classes make heavy use of the STL and templating. And to be honest, I didn't find them that handy -- they were more like someone else's idea of how to organize on a higher level, and while I'm all for "organizing on a higher level", I just didn't like how it was done.

I find that true for many formal OO API's I have to work with (generally, in perl); I'd much prefer they were more minimal, and allow me to class them up in a more custom way. OTOH, having something generic ready out of the box is less work and usually makes things simpler. But I always feel I'm am going along with someone else's plan and grind my teeth when it is not the plan I would have liked.

Currently, my preference for GTK stuff is C++, but using my own classes built from the C gtk+ API. Gtk+ uses its own types (gstrings, etc) anyway. So if you are into C++ and gtk+, they can go well together -- just stay away from gtkmm, IMO.

I am seriously considering you a troll.