attn cpp coders

ralphbsz said:
Fortunately, C code is usually clean, clear, readable and understandable. For example, it should be obvious that the following (valid) C program ...
Click to expand...
No longer valid as of GCC 14 / Clang 16 (which started enforcing C99):
www.redhat.com

New warnings and errors in Clang 16

Clang 16 will be available with Red Hat Enterprise Linux (RHEL) 8.9 and 9.3. Like Clang 15, it comes with some new warnings and errors enabled by default that more strictly enforce language standards and help prevent bugs.
www.redhat.com www.redhat.com

Porting to GCC 14 - GNU Project

gcc.gnu.org gcc.gnu.org
😞
ralphbsz said:
define _ -F<00||--F-OO--;
Click to expand...
There's a hash (#) missing at the beginning of this line

Compiling with -fpermissive I get:
Code: 
$ gcc -fpermissive foo.c 
foo.c: In function 'main':
foo.c:3:12: warning: implicit declaration of function 'F_OO' [-Wimplicit-function-declaration]
    3 | int main(){F_OO();printf("%1.3f\n",4.*-F/OO/OO);}F_OO()
      |            ^~~~
foo.c:3:19: warning: implicit declaration of function 'printf' [-Wimplicit-function-declaration]
    3 | int main(){F_OO();printf("%1.3f\n",4.*-F/OO/OO);}F_OO()
      |                   ^~~~~~
foo.c:1:1: note: include '<stdio.h>' or provide a declaration of 'printf'
  +++ |+#include <stdio.h>
    1 | #define _ -F<00||--F-OO--;
foo.c:3:19: warning: incompatible implicit declaration of built-in function 'printf' [-Wbuiltin-declaration-mismatch]
    3 | int main(){F_OO();printf("%1.3f\n",4.*-F/OO/OO);}F_OO()
      |                   ^~~~~~
foo.c:3:19: note: include '<stdio.h>' or provide a declaration of 'printf'
foo.c: At top level:
foo.c:3:50: warning: return type defaults to 'int' [-Wimplicit-int]
    3 | int main(){F_OO();printf("%1.3f\n",4.*-F/OO/OO);}F_OO()
      |                                                  ^~~~
$ ./a.out 
0.250
 
Jose said:
I think Python is gross, but I'm just a weak-minded Java programmer. I can probably be ignored safely.

I recently ran into something like this at work:
Python: 
def __getattr__(self, name: str):
        # If class is extened, or name doesn't start with query_by_
        if self.__class__ is not DatabaseTable or not name.startswith("query_by_") or not name[9:]:
            raise AttributeError(f"{name} not found")
        bin = name[9:]
        schema: DatabaseSchema = self.__dict__.get("schema")  # type: ignore
        if bin not in schema.index_keys:
            raise AttributeError(f"Index not found for attribute {bin}")

        def func(
            value: str | int | None = None,
            modifier: Callable[[Query], None] | None = None,
            policy: dict | None = None
        ) -> Iterable[DatabaseResponse]:
            return self.query(bin=bin, value=value, modifier=modifier, policy=policy)
        return func

Maybe you're smarter than me and what this does is intuitively obvious to you.
Click to expand...
Hello Jose, I am not a Python programmer and i have yet to read anything about this language. I'm assuming it is Python and i get a headache looking at this code. C, C++ and Java are more like PHP and i feel more comfortable working with these languages. However, if you want an answer, then i will post a guess as to what it is doing:

a class to check attributes of a database and its schemas?
checking an attribute string via slice starting at index 9?
containing a boolean testing the object is callable?
containing a boolean testing if the object is iterable?
error handling is self-explanatory

intelligence has nothing to do with it. One may not be able to understand someone elses code if one has not learned the language or learned the language well enough. But reading and understanding someone elses code can oftentimes be like a heavyweight fight without the gloves.

I will now post some of my dirty php code and see if you can guess what it is doing. My point is that whenever we are coding as part of a team, then we need to code with other coders in mind. obscurity is very bad. keep it clean and understandable.

Code: 
function name_is_concealed_for_this_thread($m, $n, $v, $mp, $np) {
  $mt = (array) ['d' => 'center', 't' => (array) []];
  $half = floor(($v + 1) / 2); $halves = $half + 1;
  $t1 = $half - floor($half / $m);
  $t2 = $t1 + 1; $t3 = $t1 + $n; $t4 = $t3 + 1;
  switch ($np) {
    case 0:
      if ($mp) { $mt['d'] = 'vertical'; $mt['t'][$t1] = $t1; $mt['t'][$t3] = $t3; break; }
      $mt['t'][$half] = $half; break;
    case 1:
      if ($mp) { $mt['t'][$t1] = $t1; $mt['t'][$t2] = $t2; $mt['t'][$t3] = $t3; $mt['t'][$t4] = $t4; break; }
      $mt['d'] = 'horizontal'; $mt['t'][$half] = $half; $mt['t'][$halves] = $halves; break;
  } return $mt;
}
 
Jose is the specialist here detecting how you can abuse any language. And he is right. Any language can be abused. Even ADA.
But with this thinking in mind, you will never learn any language as all can be abused. So you must make a choice.
- Start with python if you want to create something quick (& a bit dirty), as as far as i know , no static type checking, and probably a garbage collector running in the background.
- Start with C if you want to learn internals, yep, the kernel is written in C.
- Start with C++ if you need Classes e.g. for developing a GUI application using QT6.

Me , i use NIM or DLANG when I can, but then the user base is low & no professional uses it.
Both languages have compile time static type checking , NIM has reference counting I think & DLANG optional garbage collection.
And neovim-qt editor is fantastic, coloring , code completion, info about libraries,functies,types of these functions, errors as you type.
When the editor says ok, 99% chance , program compiles to a runnable executable without any error.
 
Alain De Vos said:
Jose is the specialist here detecting how you can abuse any language. And he is right. Any language can be abused. Even ADA.
But with this thinking in mind, you will never learn any language as all can be abused. So you must make a choice.
- Start with python if you want to create something quick (& a bit dirty), as as far as i know , no static type checking, and probably a garbage collector running in the background.
- Start with C if you want to learn internals, yep, the kernel is written in C.
- Start with C++ if you need Classes e.g. for developing a GUI application using QT6.

Me , i use NIM or DLANG when I can, but then the user base is low & no professional uses it.
Both languages have compile time static type checking , NIM has reference counting I think & DLANG optional garbage collection.
And neovim-qt editor is fantastic, coloring , code completion, info about libraries,functies,types of these functions, errors as you type.
When the editor says ok, 99% chance , program compiles without any error.
Click to expand...
Hello Alain De Vos, thanks for the advice:) I appreciate you taking time to reply. Also, i am somewhat aware of Jose. I have seen threads where he has posted and he is very knowledgeable. Actually, alot of members in this forum are quite capable coders :)
 
Well we can guide you to tiktok & twitter (joke).
Here a link why i personally like neovim-qt,
Alain De Vos

Post in thread 'neovim configuration.'

What does this gives ?
- To the left you see a tree of your directory and files, which you can open.
- On top you have a bar with different files you opened
- There is a fuzzy finder to find files
- On bottom there is a statusline
- To the right there is a scrollbar.

Example editing python file,
test.png
  • Like
 
kpedersen said:
I would do a quick experiment with both (C++ and Python). Make a very simple program that reads pixels in from OpenCV and displays them on a simple GUI window. This is useful because it gives you a taster of setting the project up and how dependencies will be handled for each.

  • If you find the effort comparable, then I would go with C++. As mentioned it may be a bit slower to get started with but the quality of output will be higher and more resilient.

  • If you find yourself struggling, then I would stick with Python. You don't want to get bogged down with some of the nonsense C++ creates when your purpose is to make the program itself.
Click to expand...
excellent advice. Thank you for your time and effort in this thread. I appreciate you very much.

I am going to read some more books today and then start tinkering with code.

I want to learn C++ regardless. I have much interest in continuing my education. I once played with assembler language to get an idea of how it works but i never got around to learning C or C++. Web Development has kept me chained to HTML, CSS, JavaScript and PHP. I once learned QBasic when i was using Windows 98 but Microsoft languages are disturbing in my opinion. I stay away from MS code.
 
just one note, thickness of books give an indication time needed to learn a language
- C Programming a modern approach, by K.N. King, 800 pages.
- Programming principles and Partice Using C++ , by Bjarne Stroustrup the inventor of C++ , 600 pages but multiply by a factor 2 , as it's rigorous and might need two passes to grasp the intent.
- Mastering Nim, A complete guide to the programming languge , by Andreas Rumph, 300 pages.
 
Alain De Vos i am not familiar with Nim. I've never heard of it until now.

regarding C++ books, i have the following books:
The C++ Standard Library (Nicolai M. Josuttis)
Data Structures and Algorithm Analysis in C++ (Mark Allen Weiss)
C++ Data Structures (Nell Dale)
The C++ Programming Language (Bjarne Stroustrup)
A Tour of C++ (Bjarne Stroustrup)
Programming Languages C++ ISO/IEC 14882
An Introduction to Programming with C++ (Diane Zak)
Modeling Derivatives in C++ (Justin London)
C++ for Engineers and Scientists (Gary R. Bronson)
C++ for Everyone (Cay Horstmann)
Cross-Platform Development in C++ (Syd Logan)
C++ for Mathematicians (Edward Scheinerman)
C++ Solutions for Mathematical Problems (Arun Ghosh)
C++ The Complete Reference (Herbert Schildt)
C++ 20 The Complete Guide (Nicolai M. Josuttis)
C++ How to Program (Paul Deitel, Harvey Deitel)
C++ Network Programming (Douglas C. Schmidt, Stephen D. Huston)
C++ Primer (Stanley B. Lippman, Josee Lajoie, Barbara E. Moo)
API Design for C++ (Martin Reddy)
Expert C++ (Marcelo Guerra Hahn, Araks Tigranyan, John Asatryan, Vardan Grigoryan, Shunguang Wu)
Effective Modern C++ (Scott Meyers)
 
Haha, just like me but worse , for me it would be,
1. C++ primer
2. Effective modern C++
3. Datastructures and Algoritms.

The rest for the library, reference , too look, when i hit something.
 
Most of the time, just use std::vector. It beats everything else hands down most of the time. Quite often you can use std::vector instead of any of the map/set containers (typically when you have an integral index that is contiguous). std::flat_map and std::flat_set are also good choices.

You need to take care of std::vector insertion resulting in iterator invalidation.

When the key is either non-integral or there are large holes in the range then you will need the map/set containers.

You will rarely or never need to use deque, stack, queue or priority queue. The performance of std::list is so bad that you will only ever need it if you are splicing many large objects into the middle or beginning of the container.

I haven't played with it yet, but C++26 added std::hive which should have most of the benefits of std::list (stable iterators and lower cost insertion) without the large memory overhead, spending vast amounts of time in new/delete and chronic memory fragmentation.
 
Paul Floyd points out something very important in his last sentence:
memory stuff. There can be some hidden copies and such generated resulting in new/delete that you don't realize and can have hidden side effects.
For stuff in the STL make sure you read the documentation carefully, pay attention to "is this threadsafe" if you are doing threading.
 
Paul Floyd said:
Most of the time, just use std::vector. It beats everything else hands down most of the time. Quite often you can use std::vector instead of any of the map/set containers (typically when you have an integral index that is contiguous). std::flat_map and std::flat_set are also good choices.

You need to take care of std::vector insertion resulting in iterator invalidation.

When the key is either non-integral or there are large holes in the range then you will need the map/set containers.

You will rarely or never need to use deque, stack, queue or priority queue. The performance of std::list is so bad that you will only ever need it if you are splicing many large objects into the middle or beginning of the container.

I haven't played with it yet, but C++26 added std::hive which should have most of the benefits of std::list (stable iterators and lower cost insertion) without the large memory overhead, spending vast amounts of time in new/delete and chronic memory fragmentation.
Click to expand...
Excuse me if i use the wrong wording. But there is a tree of data-structures. And Scala did it very fine. But one does not want to use Kotlin. Sorry i know nothing about C++ , but there is always dynamic & static. Meaning it can grow or it can grow not.
And there is you can put in anything you want , or it must always be off the same time. This makes always speed &
efficiency tradeoffs.
 
Speaking of different collection classes, here is a psychological test whether you are "compatible" with C++:


This introduces a new collection class. It does clever things to make a collection class with a mix of efficient operations that wasn't in C++ before. This introduction is very long, although the class introduced is not overly complicated. It just does a very complete job of specifying every detail and "selling" it, mainly to the C++ standards committee members.

Can you stand this? C++ will be a lot of reading, and be told about a lot of details. As a result you will be able to write more elegant programs and faster code, sometimes both at once. Is that what you want?
 
Alain De Vos said:
just one note, thickness of books give an indication time needed to learn a language
- C Programming a modern approach, by K.N. King, 800 pages.
- Programming principles and Partice Using C++ , by Bjarne Stroustrup the inventor of C++ , 600 pages but multiply by a factor 2 , as it's rigorous and might need two passes to grasp the intent.
- Mastering Nim, A complete guide to the programming languge , by Andreas Rumph, 300 pages.
Click to expand...
R5RS Scheme specification is 50 pages.
 
cracauer@ said:
Speaking of different collection classes, here is a psychological test whether you are "compatible" with C++:

Click to expand...

I think that the final version of that paper is https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2024/p0447r28.html

For those of you not familiar with C++ standardisation, note that this is r28. The 29th version (counting starts from 0). The initial version was back in 2016 https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0447r0.html when it was called std::colony.

The paper was too later for C++17 and wasn't accepted for C++20 either. It got de-platformed for C++23 so the author wrote paper to try to get it back on the agenda. https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p2523r0.html

During ratification by National Bodies the question of constexpr hives came up. Not sure where that went.
 
johnjohn said:
Alain De Vos i am not familiar with Nim. I've never heard of it until now.

regarding C++ books, i have the following books:
The C++ Standard Library (Nicolai M. Josuttis)
Data Structures and Algorithm Analysis in C++ (Mark Allen Weiss)
C++ Data Structures (Nell Dale)
The C++ Programming Language (Bjarne Stroustrup)
A Tour of C++ (Bjarne Stroustrup)
Programming Languages C++ ISO/IEC 14882
An Introduction to Programming with C++ (Diane Zak)
Modeling Derivatives in C++ (Justin London)
C++ for Engineers and Scientists (Gary R. Bronson)
C++ for Everyone (Cay Horstmann)
Cross-Platform Development in C++ (Syd Logan)
C++ for Mathematicians (Edward Scheinerman)
C++ Solutions for Mathematical Problems (Arun Ghosh)
C++ The Complete Reference (Herbert Schildt)
C++ 20 The Complete Guide (Nicolai M. Josuttis)
C++ How to Program (Paul Deitel, Harvey Deitel)
C++ Network Programming (Douglas C. Schmidt, Stephen D. Huston)
C++ Primer (Stanley B. Lippman, Josee Lajoie, Barbara E. Moo)
API Design for C++ (Martin Reddy)
Expert C++ (Marcelo Guerra Hahn, Araks Tigranyan, John Asatryan, Vardan Grigoryan, Shunguang Wu)
Effective Modern C++ (Scott Meyers)
Click to expand...
This is like collecting & (may be) reading a whole bunch of books on carpentry, instead of spending the bulk of one's time wood working and building stuff! Programming has a lot more in common with carpentry than mathematics. Building stuff, making mistakes and learning from them is part of the process. You can read books *later*, to organize what you learned. Embrace the phrase "learning by doing"!

Also, programming languages are just tools. Better to learn more than one and not fixate on "one true language".
 
You must log in or register to reply here.
Back
Top