Master Pointers in C: 10X Your C Coding! 

Sounds like IBM mainframe HLASM. I've been doing that for a little over 40 years. Although I vaguely remember some 8 bit computers which had an index register. I don't know x86.

I used indirect addressing in assembler for some years before I realised I was using pointers, when I did a Pascal course around 1984.

I would have struggled less in the beginning if someone would have told me that a pointer is like 6502 zero page!

​@@DavesGarageI programmed the 6502 way back in the 80s, but I didn't have an assembler. I did everything directly in hex op codes etc. But that stuff definitely helped me later with pointers in C i.e. they're just memory addresses as you said.

Dad's should teach assembly language to their children. I think of it like the Latin of computer languages. You learn about how the computer actually works and appreciate all of the things that come along for free with the HLLs that you will use. You also see pointers in things like the aptly named Stack Pointer and the Instruction Pointer. You also see function pointers in vector tables. Assembly is cumbersome but you learn so much by doing just a little bit.

I'm 12 years into programming now... And i just understood the concept now, with that analoy :D

as if you never heard that analogy before. that analogy was in every tutorial I've seen.

Hope it helps others!

What kind of statement is this? 😂

@@itsSpidey- he is saying that dave's appearance exudes competence and advanced knowledge of the subject

Haha, all programmers are lazy; this is what drives the development of technologies, languages and so on!

@@tangiblewaves3581 well i agree but i dont agree, i think we always find shorter ways to do things and yes we develop wicked tech from it, but programming its self can be very intensive / labor intensive. And takes a lat of work even to learn, so i get what your saying, but programmers are just like any trade (plumbers, electricians) and just like those trades look for better ways. I have a family full of trades, and they are full of unique (what i would call lazy) ways to fix stuff but it works and they have good solid reputations for the work they do, sometimes its those ways that make a completely new product that works 10x what the previous method is, but it takes knowledge and the skill to think it through (and sometimes the balls to apply the fix lol and hope it works.) just like them, programmers, designers and scientist need to sometimes "go back to simple" in order to step forward. All that said... i have seen lazy programmers and lazy plumbers. so i think the ratio is likely the same in terms of avoidance of physical labor. o7

@@theMaizeOfficial exactly, and if you want to code proffessional, you also need some good amount of discipline to code even when you have a bad day or are tired.

Cool story.

A good programmer is that who will spend 3 hour of his time to automate a one time task that requires half an hour to be made manually.

Assembly is a lost art, but explains so much.

this

Yeah assembly could have explained it simply.. You have Registers, and direct/indirect addressing… all registers can be pointers unless they all are general purpose registers Then the addressing and instructions decide how we are representing the registers LA R1,=A(0000) Load address of location where =A(0000) has been resolved in the executable, into Register 1 XC 4(256, R1), 4(R1) XOR 256 bytes starting from values pointed by R1 offset by +4 bytes with the same… (XOR of itself is 0 ie clearing all the bits/bytes at that location) ST R1,INT32 Store value in R1 into location INT32 L R3,INT32 Load value at INT32 into R3 LA R4,INT32 Load absolute address of symbol INT32 into R4 ST R3, 0(R4) Store R3 into INT32 symbol location INT32 DS 4F Someone who knows assembly should be able to teach pointers in a better way

"having played around with languages that didn't have dynamic allocation" Cobol ?

@@IBelieveInCode Logo and Visual Basic, though I was very young at the time and no internet so I might just not have been aware if they actually did have that capability.

Honestly i don't have any problems at all understanding the concept of pointers, all that makes sense to me. What i don't understand, is when you want to use them

Maybe that is the right way to tech the concept, is to pose problems that require the indirection to do the job. A lot of things are easy if the person thinks it was their own idea or something they want rather than something imposed from outside. Get them to say for themselves "what I really need is some way to ..."

@@TehIdiotOne as said on this video, the most common use is with larger or variable data types like objects, it is much more efficient to point to the adress of an object than bring the entire object into memory, this is especially noteworthy in C++ when storing lists of large objects or integers, you just use the pointer and the -> operator and it is way more memory efficient. Plus, it allows for dynamic memory allocation which allows to manage what is and what isn't in memory at any given time

int* should be the only way to write int pointer.. everything else should be banned 😄😄 To me it makes the most sense. The type is an "int pointer" so put it with the type name... I think it's confusing for beginners because it looks similar to the dereferencing syntax.

@@JamieT1986 yes, but then a declaration in 4:07 would create 2 pointers, not one. int * x, y is in actuality: int *x AND int y. I agree it would be better to just have something like 'intP x, int y' and then it would be obvious what is a pointer and what isn't but we got stuck with a syntax that is efficient for writing, but hard for reading - somewhat opposite to Python's syntax.

@@vaakdemandante8772 int* x, y; is something I was a little skeptical about actually when I saw this. I never knew that's how it worked, and I forgot to test it myself. I will assume it's correct... but yeah that kinda screws it up! :)

"int \*num" is much better, the declaration in 4:07 is confusing if you use the "type\* var" version. Edit: I used "\" sign cuz youtube comments converts asterisk to bold letters. lol

This^^^ i would say, find a style and stick to it, but as an OS and firmware dev the most common style i come across (and use personally) is this: Type* var_name = ….whatever here…. Ex: int* pointer_to_integer = (int*)malloc(sizeof(int)); So the asterisk essentially lives right next to the type, with no whitespace between. This makes it clear that it is a POINTER of said type, in my example a pointer to an ‘int’ Using this style consistently makes reading and reasoning about pointers much less frustrating in my opinion. Of course, then there is dereferencing a pointer to get the object’s value. In those cases common style i see and use is this, continuing with my int example: int integer_variable = *pointer_to_integer; So we use the asterisk right against the VARIABLE NAME to dereference it. The combination of sticking to these two basic style guidelines makes points way less messy. Essentially, use the asterisk directly next to your type name when declaring a pointer variable, and right next to the variable name when de-referencing it. Hope this helps some of the less experienced! Also, as a follow up, i think the example at 4:07 of Int* x, y Is a kind of poor example. It just isnt code that would fly in most (well maintained) C projects like the linux kernel for example. You want things to be CLEAR above all else, and so its not really much effort to be explicit with each variable declaration. The compiler can optimize for you if it is necessary or possible. Otherwise, be more verbose in an effort for future contributors to have an easier time grasping what your code is doing. Little syntax tricks like that can be neat, but often fall flat in well maintained project environments with C.

Making a drawing of it often helps a lot.

Hang in there! You'll understand them 👍

C and C++ learning is like this: you learn the basics and start writing. Then you want to do more or understand someone else's code, at that point you go through 3 to 6 months of FOG where you don't quite get it, but you keep trying. The good news is once you get past that, you become a decent programmer.

@@newmonengineering I basically set out to make a reasonably simple game and by using reference code from others or books I've managed to piece things together bit by bit. By interpreting compiler errors I've managed to kind of get a feeling for when it's in need of a pointer or not. So I definitely know more now than half a year ago luckily :)

I think C started a bad pattern for working with pointers because you declare one with the same symbol you use to dereference one. That said the arrow operator has some, as of yet untapped, potential for conveying ownership semantics needed for automated resource management (at the cost of the control C devs freak over :p)

We are that old, eh?

I first used pointers in Pascal on a mini computer in the mid 80s, and then Turbo Pascal in 1986, although I was also an assembler programmer before that, where you use pointers all the time but call it indirect addressing. Pascal taught me about structures. Then I went onto C, but it was K&R, and I found the nomenclature hard to follow: it took me months to get proficient, not because I didn't understand the concepts, but because K&R C can be super impenetrable for a C noob. Thankfully ANSI C came along at just the right time, I re-wrote the K&R code I was maintaining in ANSI, and I've stuck with that ever since. C remains my language of choice to this day, but mostly I'm a deeply embedded systems programmer and mixed signal hardware engineer, where C sits well, nice and close to the hardware. I dip into assembler still every now and again for those pinch points, but 99% is C.

​@nezbrun872 I literally use Delphi for fun nowadays, with a free Community license. It's pretty fun tbh

Glad it was helpful!

"Usually when we program in C without classes ,we are using malloc" It depends on the platform and application: in embedded microcontroller applications, which are typically memory constrained. would fail catastrophically in the event of failure, so malloc is rarely used in production hardware. As a result, memory and objects are typically allocated statically, or on the stack (typically small objects only on the stack).

Calling C++ “C with Classes” is sort of like calling Firefox “Netscape Navigator”. Or maybe like calling Windows “MS-DOS 1.0”.

@@danielrhouck It is something common among C (ansi, K&R, 90, etc) fans.

@@JohnDoe-kh3hy Classes are one of the *many* things C++ does differently than C, and they are *not* the most important one to this discussion.

​@@danielrhouckNot really. Stroustrup referred to C++ as "C with classes". If it's good enough for him, it's good enough for me. It's a nice succinct way to describe the language conceptually.

You're right, I remember "struct IntuitionBase * pIntuitionBase = (struct IntuitionBase *) LoadLibrary(:...)" is what threw me the first time!

@@DavesGarage We really needed the modern version of the auto keyword back on the Amiga. 80 columns only go so far.

Because i guess you did not start with assembly langauge. This was my 2nd language after Basic .... like all teenagers in the early 1980s home computer time. Feeling sad for the current generation that think it's not only magic but black magic.

@@llothar68 actually, in my Electrical Engineering curriculum in 1999 I took a class in ANSI C in Unix and an Assembly class. In the Assembly class I wrote a 8088 CPU emulator, which was both frustrating and fun! I ended up using assembly programming PIC microcontrollers more than anything else. I used the term “magic” because I know what it is doing under the hood for me, and how much effort the C language saves us rather than writing in pure assembly! Cheers! Edit: it was the 8080, not the 8088-long time ago and I have lost the floppy disk where I had a backup. Oh well…. My degree is in Electrical and Computer Engineering (ECE) and I took quite a bit of digital design classes, focusing more on hardware than software.

@@llothar68 im self taugh software engineer, i did not went to cs, and I know this stuff, cause I love digging deep, no need to feel sorry. most of these jobs doesnt require this knowledge, yo ucan still learn it as hobby

Also too, (kept in my mind) the stack on 6502 is so small you could never pass a struct like item to a function with anything but a pointer.

Function pointers are kindof the origin of object oriented programming. You would create a struct with function pointers as member variables. Then a "constructor-like" function could initialize the struct with the desired functions and voilá you have a class. C++ started out as a transpiler to C and, I might be wrong on this part, that is how i think it translated c++ classes into c structs.

Function pointers are currently almost not used (directly) in modern c++ (of course they exist and can be used if needed): most of the functionality was moved into lamba/function objects which is more robust and powerful solution (with some small problems). Also operators new/delete should not be directly used (as smart pointers are better option with make_shared and make_unique templates).