Thanks for watching! Hope you enjoyed the video! Don’t forget you can try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/TheCherno. The first 200 of you will get 20% off Brilliant’s annual premium subscription!
1. Thanks for this video. It makes a lot of sense and a good example for newbies on how to structure a C++ project properly. 2. At this point, on Dec 31 of 2023, this is the last video in the "C++" playlist. And I watched every single video from it. Which took 23+ hours in total + half of it for writing down notes. THANK YOU for making this whole RUclips-based course. 3. Yeyyyyy!!!
Right? Gonna get 4,000 hours of quality YT instruction from this one guy's 2000 hours of development. Hell, we got 45 minutes from the Readme lol. Love it!
If you want to show us how you'd refactor his code from a maintainability standpoint into a proper setup, I think that would be extremely useful. Too often we focus on refactor for performance, but in this case, just not having to edit two files for one change is a great lesson to impart.
Give up with this “proper” nonsense. There’s is no proper way to do it. Gee a lot of programmers are dumb, and can’t comprehend that their way is not the “proper” way, but simply one of many.
Proper doesn't mean "only". There are many correct ways to do it. What's being argued for here are better ways to do things. In this case, empirically, as making a small change to a class required updating the same code in multiple files. I disagree, there are definitely proper way(S!) to do things, and ways that one should absolutely not. @@darmath2009
From what I understand, CMake also generates projects for Visual Studio. In this case, it's cross-platform, creating a Makefile on Unix and a Visual Studio project on Windows
More than that. Several IDEs just support CMake out of the box (i.e. they run the necessary CMake command for the given build environment internally and support the entire thing in the IDE, even helping you when you add files, no need to run any batch or shell scripts...). On top of that, you got things like vcpkg which is a package manager centered around CMake (and MSBuild) that can help you a lot. Got some project with a CMakeLists.txt? Just open the folder in VS Code (with proper extensions), Visual Studio, CLion, ... that's all. A bunch of years ago, we we're still using some batch script or shell script that would generate solution files or makefiles, but then support was added in Visual Studio, CLion also supported it (i think even before VS) and then VS Code came along with its platforms which can be used as an IDE supporting CMake projects on windows, linux and mac. Since CMake is used so widely, it's trivial to get any CMake library setup to be able to be used in your project, even if it is not contained in your package manager.
I designed something like this for my personal projects. No specific core library, but several other libraries which can be picked from and offer a little more freedom to minimize app footprints. Separating apps and libraries into corresponding subdirectories can help with organization and reusability
At this point, I watched Cherno videos not only to learn but also as a radio, it makes me feels smarter watching your videos in the background while working XD
Nice to see a video about project setup. But please people, prefer out-of-source builds! CMake supports and recommends it, I hope Premake also. While having your generated project files sitting next to your code sounds like a good idea, it will bring more problems than solutions in the future. Build into a separate folder, you can even have different folders for different build types or experiments. Deleting a single folder ensures a clean state of the project. in CMake: git clone cd mkdir build cd build cmake .. cmake --build .
Absolutely! I think that in case of crossplatform development and in case you want to make such components management the best solution is git-modules and cmake. cmake works on any platform you like, it can generate *.sln file for Windows and xcodeproject in case of MacOS. Cmake has special folders management called include(GNUInstallDirs). And it is much more pleasant then pure MAKE of unix
A quick look at the docs and you should be able to do something like this untested code: ``` newoption { trigger = "workspacedir" value = "path" description = "Set the build directory, defaults to current directory" default = "." } assert(_MAIN_SCRIPT_DIR == _OPTIONS["workspacedir"], "Don't do in-source builds!") workspace "MyProject" location(_OPTIONS["workspacedir"]) filter { "configurations:Debug" } targetdir(path.join(_OPTIONS["workspacedir"], "build_debug") filter { "configurations:Release" } targetdir(path.join(_OPTIONS["workspacedir"], "build") ```
@@CoDEmanX47 I guess technically you're right 😆 but I hope it illustrates the point, you can create the folder wherever you feel like. Which is what IDEs that support CMake do many times, creating the folder somewhere else.
@@michaelterekhov1678 I agree except that git submodules are a pain in the ass unless the submodules are all completely stable. It's not something I would recommend if you're still actively working on the modules as well as the main application. You could do it as a library in a sub-folder and move it to submodule when it stabilises.
It's 2024 and people are using Lua to build a C++ Project to implement a Library in Lua that helps them to easier build of C++ Projects via Lua. Pretty good reason to exist.
Just a point of clarity on build systems: Both Premake and CMake are build system generators - all of the benefits that you enumerated for Premake apply to CMake as well. It would be good to see how you pull in dependencies to this template as well - do you do it manually or with a package manager?
Depending on what platforms you ship to, using Visual Studio solutions can be fine. You don‘t have to use a build system if you don‘t need it. Commercial games and engines have been shipped using just native IDE solutions. Setting up a buildsystem however is a lot more important for open source software, where you probably want to run it in as many places as possible. So consider if you need it, what benefits it brings, and if it‘s right for you.
Couple of items. Even watching here, we see repeated common blocks in your premake lua files (configurations, preprocessor defines, compiler settings, directory patterns, architectures...). I've found it very useful having a file "common.lua" and executing that using "dofile" in each premake file. Don't Repeat Yourself is a good goal. Second, is imGUI actually embedded in the source? Might be good time to discuss git submodules.
No, it always depends on the team and the field you’re working in. For example, in embedded/gamedev, rarely does someone use a package manager because most of the time, you implement stuff from scratch rather than using some external library.
Agree. Now that vcpkg has manifests, it fixes a giant hole that previously you'd have needed conan for. I'd been using vcpkg instead of conan even before that, just preferred it, but I ran into issues where I'd check out an old repo on a new vcpkg install and the newer package versions broke it. Now you can specify dependencies and version requirements in a manifest and use it like a reasonable package manager (ofc, conan has had that long before).
Before I used Autotools it was simple to use, but it was limited to Linux build systems, since I expanded my projects to windows and, I migrated the build system to CMake
Dist is what most IDEs and people refer to as Release, I.e. a configuration of full optimizations and no debug information, whereas Release is a Dist build with debug information
@@marcasrealaccount That... doesn't make sense. "Release" should be what you actually, you know, RELEASE into the world, to the public. Adding debug information to release is like... bro, what are you even doing?
Hey @The Cherno so I followed along your video and used this template. I included GLFW and GLEW into the Core project (as much of the core engine is using these libraries), and I linked the two libraries in the Editor and Runtime projects. The result is a working game engine, but the console is spammed with a bunch of link warning 4217. I'm not one to disable warnings, but the project is running smoothly. After some research I've come to the conclusion that since the Core project (a static library) is including the header files of GLEW and GLFW but not the implementations, when the compiler tries to replace all the function calls from GLEW or GLFW in the Core project files, they're using symbol names (such as "__glewAttachShader") instead of their corresponding function names which get replaced in the Editor and Runtime projects. Is there a work-around to this? Or a way to fix these warnings?
No offense, but the guy could have used 10 hours of those 2000 hours to learn about design and save himself from all the trouble of duplicating a whole class just to add editor tools to it.
Damn your so right, but 10h wouldnt be enough time to learn CMake or premake and actuallz apply these core-app concept. I srated this project when i was 15 and learning back then was harder compared to today using ChatGPT (its literally a cheay). I have spent the last week 60hours tlearning CMake and refactoring the whole engine into a core app structure, I LOVE. IT. im so glad i sent him mz project
How come do you go to the end of the line and add a semicolon so fast? I'm looking everywhere if is there any shortcut to jump to the end of the line. People often suggest Shift+Enter, but it doesn't work.
I'm new to this now I'm here in this c++ playlist & i know how to make projects with unreal engine, good but still understand nothing from this should i watch the game engine playliist to understand what is in this video, learn get or where to go now.
So i was just setting up this template to be used with glfw where i wanted to wrap glfw in the core and only expose a non glfw specific api for the app. One thing i noticed was i had to include the glfw lib as an include of app as well if i included my wrapper even know the wrapper doesn't expose any glfw specific types. Seemingly i should be able to compile glfw into my "core" lib then only include it in the app project right...? Did i set something up wrong or is this not the case?
Generally a bad idea to teach premake where 90% of all projects are cmake based in the c++ world and the rest are either make or .bat/.sh scripts. I have never seen any major library be built with premake so i would highly recommend investing your time into learning cmake over premake as it really is the industry standard besides its pain points that the cherno has. Bad take honestly on this one
Highly disagree. Cmake is an ancient relic that should be deprecated in any sensible codebase. It‘s not backward compatible, causes many points of breakage, is slow, has terrible documentation and generally causes more issues than it solves.
I do agree with it although premake is lot easier than Cmake in my opinion, also for gamedev it's pretty easy to setup. I just spent 3 days to fully setup a pretty complex HPC based library for my next project in Cmake so I am pretty damn salty because of all the different compilers I have to run and flags I support.
CMake is literally trash. I started with using CMake and it's so confusing, has bad or non existing learning resources and is so complex without any reason. Premake is very beginner friendly and is highly configurable at the same time
I totally agree. If you lack knowledge of any preprocessor or project configuration tool, you are NEVER going to create a complex enough project to need the "simplicity" of Premake over the much faster CMake. CMake is much more difficult to configure? True, but only for large code projects. For simple or even medium size projects, CMake is the way to go, and one of the key points is that is the industry standard. DO NOT start with Premake for a simple project, you will spend more time setting all the files and configurations than just running cmake .. && make
I have the same question. Modern CMake is easy (with the correct docs) and straight forward: target_something. Precompiled headers, done. Unity builds, done. Packages, done (conan or Fetch Content). Pre or post build scripts, done.
Hi Cherno. Why is Project Red (Cyberpunk ) switching from the RedEngine to Unreal5 ? Why not just improve on their own engine? Isnt making games with your own engine more profitable on the long run? Is it because of some tech like Lumen and Nanite that coders at Project Red can not replicate?
I have a project were i need to use clion. Premake has no good support for clion. Clion does use cmake which premake can genrate but clion overrides some cmake stuff and i had to make my own version of premake-clion with some ugly hacks. No need to say i switched to cmake in the end.
I used to be on the Premake train. But the integration with IDEs that CMake has just does not compare at all to what Premake offers. Premake is certainly easier for a primarily Visual Studio user for sure. But if I want to develop on Linux and not just "build" on Linux for instance, I want CMake. I'm not some die hard VIM user, I want to be able to use a graphical IDE anywhere.
Cool looking template. What do you usually use to manage dependencies? I recently started using conan for basically all C++ projects, especially when things like OpenGL, ImGUI, etc. are involved.
I think I tried conan before, if I remember right, it has a CMake integration too. I myself use vcpkg (both with CMake, as well as in situations where I am working with visual studio solutions, since it works there too). VCPKG has all the stuff you'd need to OpenGL too, including imgui (with configurable bindings, i.e. imgui[glfw-binding,opengl3-binding] if you want opengl plus glfw setup, which is what I always use). You can also use it as a submodule if you want, since it is itself just compiled with C++, it's easy to set even that up cross-platform in your CMake script so that someone doesn't even have to manually install it. vcpkg also recently added manifests, which was much needed (i.e., specifying dependencies in a vcpkg.json file in your project, along with version requirements, just like conan does it)
What a mess it can be if not structured properly from the beginning. This is when we would have needed a private Cherno-GPT that did what Cherno does here, and more. Well. Cherno-GPT sounds nice.
initia*LIED,* yeah. 😂 No wonder it couldn't find it. I have a few of those in my own engine. 😳 Would've been nice if Visual Studio was checking some basic grammar (and if you clicked on it, rememberd that you actually WANTED it misspelled). 😩
Say if I have a game engine that has its own renderer and physics (engine) and some other components/engines (audio, animation, etc). Are these components/engines the "Core" and then my Editor or a game can then be the "Apps" and they can include/use the core stuff? Sorry if this is hard to read and understand and my question is that does this follow Core/App architecture? and is that the genreal or recommended approach? If not please explain. Thanks
Hello, its me the creator of this engine haha. I spent the last week 60hours rewriting my whole engine into this core app strucuture, and learning cmake, and as far as i understood yes you are right, it should be implemented this way so you can #include from the static library into thr editor and game, and i LOVE it, no duped files anymore, thanks cherno lol
@@y4ni608 Ok. Thanks a lot for the reply. I’m also trying to do the same thing where I have my renderer as core and then I will try to include it wherever i need (editor, game, etc). But I’m actually just starting out. Wish me luck 😅
Any integration with linux IDE/Editors? Ive been looking for an alternative to msbuild since ive migrated to linux. Been using CMake and so far it's okay
how do you start adding your own files? i right-clicked the Core project in the Solution Explorer, and i added one new file in the same directory as Core.cpp and Core.h its called Logger.h i included it in Core.cpp like so; #include DEVELOPING ANYWHERE ELSE this would be fine, but because Visual Studio is beyond annoying, it will not include it and is throwing errors that it cannot find a file name "Logger.h" after pressing F5 i want to throw Visual Studio into a volcano every time i use it
The way I figured out: 1. Right-click the "Core" project -> Choose "Add" -> "New Class..." 2. Then type the name of my class "Foo" or something. 3. Add "Source/Core" to the front of both the Foo.h and Foo.cpp file path input fields. DONT use your file browser to choose the "{Solution}/Core/Source/Core" folder. It does not work. Because... reasons? By default the files just go into the root of that solution, again because... reasons? 4. Click "Ok". 5. Right click "Core" solution again and click "Build". 6. You _should_ then be able to `#include "Core/Foo.h"` in "App.cpp", build it, run it, etc. I dOn'T kNoW wHy YoU'rE hAvInG tRoUbLe? \sarcasm incase not obvious.
That's nice, I refuse to use modern IDE's still because I'm insane, so I've only really written my own Makefiles so far. Ah the pain. I think I'm going to stop making my own Makefiles now.
Im doing same architecture in Java im creating maven jar projects(core-dist-version(this is also pom)) under parent pom project for version independent projects.
One thing I don't understand in the new stuff, why do they use configuration files that have to be manually typed? Doing that you have to read the documentation. We are in 2023, can't they make a simple GUI to enable us to set everything and not to worry about typing something wrong? (not only premake, even in vscode, many extensions do that)
Hay not sure if have seen this, but have you had a look at the build chain and tooling that is coming with zig to compile/cross-compile/manage dependencies for C/C++ projects? Think it might make life easier in some way for you.
Thanks for watching! Hope you enjoyed the video!
Don’t forget you can try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/TheCherno. The first 200 of you will get 20% off Brilliant’s annual premium subscription!
This video fills a knowledge gap that I think a lot of beginner to intermediate C++ developers likely have. Thanks Yan!
1. Thanks for this video. It makes a lot of sense and a good example for newbies on how to structure a C++ project properly.
2. At this point, on Dec 31 of 2023, this is the last video in the "C++" playlist. And I watched every single video from it. Which took 23+ hours in total + half of it for writing down notes.
THANK YOU for making this whole RUclips-based course.
3. Yeyyyyy!!!
This 2000 game engine review turns out to be supper useful, thanks a lot for both of you guys
Right? Gonna get 4,000 hours of quality YT instruction from this one guy's 2000 hours of development. Hell, we got 45 minutes from the Readme lol. Love it!
If you want to show us how you'd refactor his code from a maintainability standpoint into a proper setup, I think that would be extremely useful. Too often we focus on refactor for performance, but in this case, just not having to edit two files for one change is a great lesson to impart.
Give up with this “proper” nonsense. There’s is no proper way to do it. Gee a lot of programmers are dumb, and can’t comprehend that their way is not the “proper” way, but simply one of many.
Proper doesn't mean "only". There are many correct ways to do it. What's being argued for here are better ways to do things. In this case, empirically, as making a small change to a class required updating the same code in multiple files. I disagree, there are definitely proper way(S!) to do things, and ways that one should absolutely not.
@@darmath2009
@@darmath2009 anything which saves time and redundancy is a MORE proper way.
@@RavisLifeCircle obviously English isn’t your strong point
@@darmath2009 cauz its not my first language not even second so cut the toxicity bruh
This might possibly be the most needed video for C++ developers.
From what I understand, CMake also generates projects for Visual Studio. In this case, it's cross-platform, creating a Makefile on Unix and a Visual Studio project on Windows
Yeah, CMake can produce files for several IDEs/editors and cross-platform, that's why it's so popular.
More than that. Several IDEs just support CMake out of the box (i.e. they run the necessary CMake command for the given build environment internally and support the entire thing in the IDE, even helping you when you add files, no need to run any batch or shell scripts...). On top of that, you got things like vcpkg which is a package manager centered around CMake (and MSBuild) that can help you a lot.
Got some project with a CMakeLists.txt? Just open the folder in VS Code (with proper extensions), Visual Studio, CLion, ... that's all. A bunch of years ago, we we're still using some batch script or shell script that would generate solution files or makefiles, but then support was added in Visual Studio, CLion also supported it (i think even before VS) and then VS Code came along with its platforms which can be used as an IDE supporting CMake projects on windows, linux and mac.
Since CMake is used so widely, it's trivial to get any CMake library setup to be able to be used in your project, even if it is not contained in your package manager.
Thanks!
E
I designed something like this for my personal projects. No specific core library, but several other libraries which can be picked from and offer a little more freedom to minimize app footprints. Separating apps and libraries into corresponding subdirectories can help with organization and reusability
I am impressed at how fast you can talk and explain things. I had to slow the video down a bit to keep up :).
Easily one of your best videos to date. Thanks.
I would prefere "Modern CMake" over premake and have conan or vcpkg to manage dependencies.
I recently starting using Cmake with VCpkg and my God it's the best thing that happened in a while for me. It just works like magic.
Same. But its always interesting to see alternatives.
Same... and I usually push this combo at work.
Agree - this is the way at the moment
Cmake is godawful though, "modern" or no. Not that anyone can agree on what "modern" cmake even means.
OMG, this is something I've been looking for a long time, and no one had this as a whole guide showing best practices. Thank you so much!
The video we needed from ages
At this point, I watched Cherno videos not only to learn but also as a radio, it makes me feels smarter watching your videos in the background while working XD
I was actually waiting for this for the longest time. Thanks Yan.
Nice to see a video about project setup. But please people, prefer out-of-source builds! CMake supports and recommends it, I hope Premake also. While having your generated project files sitting next to your code sounds like a good idea, it will bring more problems than solutions in the future. Build into a separate folder, you can even have different folders for different build types or experiments. Deleting a single folder ensures a clean state of the project.
in CMake:
git clone
cd
mkdir build
cd build
cmake ..
cmake --build .
Absolutely! I think that in case of crossplatform development and in case you want to make such components management the best solution is git-modules and cmake. cmake works on any platform you like, it can generate *.sln file for Windows and xcodeproject in case of MacOS. Cmake has special folders management called include(GNUInstallDirs). And it is much more pleasant then pure MAKE of unix
A quick look at the docs and you should be able to do something like this untested code:
```
newoption {
trigger = "workspacedir"
value = "path"
description = "Set the build directory, defaults to current directory"
default = "."
}
assert(_MAIN_SCRIPT_DIR == _OPTIONS["workspacedir"], "Don't do in-source builds!")
workspace "MyProject"
location(_OPTIONS["workspacedir"])
filter { "configurations:Debug" }
targetdir(path.join(_OPTIONS["workspacedir"], "build_debug")
filter { "configurations:Release" }
targetdir(path.join(_OPTIONS["workspacedir"], "build")
```
Technically, this is not out-of-source but adjacent-to-source 😋
@@CoDEmanX47 I guess technically you're right 😆 but I hope it illustrates the point, you can create the folder wherever you feel like. Which is what IDEs that support CMake do many times, creating the folder somewhere else.
@@michaelterekhov1678 I agree except that git submodules are a pain in the ass unless the submodules are all completely stable. It's not something I would recommend if you're still actively working on the modules as well as the main application. You could do it as a library in a sub-folder and move it to submodule when it stabilises.
We all waiting for your video ,Love the way you teach us ❤
This is an epic starter template!
What an underrated channel! This guy is gold! He would deserve at least 1M sub!
elite video, this type of stuff is really good for beginners
Not sure why, but I have always had a hard time with this. Thank you!
Love this guy gives so much accurate details.
It's 2024 and people are using Lua to build a C++ Project to implement a Library in Lua that helps them to easier build of C++ Projects via Lua. Pretty good reason to exist.
Thank you, I needed this!
This is an awsome video.
Thank you so much for making this video!
thankssssssssss finally I waited so long for this vid!!
I love the approach; I will definitely try it out! Thank you! 👍😉
An awesome one! Thank you so much!
Just a point of clarity on build systems: Both Premake and CMake are build system generators - all of the benefits that you enumerated for Premake apply to CMake as well. It would be good to see how you pull in dependencies to this template as well - do you do it manually or with a package manager?
The project in this video uses SFML and IMGui, and I think he added them manually.
Depending on what platforms you ship to, using Visual Studio solutions can be fine. You don‘t have to use a build system if you don‘t need it.
Commercial games and engines have been shipped using just native IDE solutions.
Setting up a buildsystem however is a lot more important for open source software, where you probably want to run it in as many places as possible.
So consider if you need it, what benefits it brings, and if it‘s right for you.
simple yet powerful, great stuff!
Couple of items. Even watching here, we see repeated common blocks in your premake lua files (configurations, preprocessor defines, compiler settings, directory patterns, architectures...). I've found it very useful having a file "common.lua" and executing that using "dofile" in each premake file. Don't Repeat Yourself is a good goal. Second, is imGUI actually embedded in the source? Might be good time to discuss git submodules.
That is especially useful for things that must always be the same across those files.
Great template! I still prefer my personal favorite combo: CMake with Vcpkg/Conan for handling dependencies.
Is not that the Industry Standart nowadays ?
No, it always depends on the team and the field you’re working in. For example, in embedded/gamedev, rarely does someone use a package manager because most of the time, you implement stuff from scratch rather than using some external library.
Agree. Now that vcpkg has manifests, it fixes a giant hole that previously you'd have needed conan for. I'd been using vcpkg instead of conan even before that, just preferred it, but I ran into issues where I'd check out an old repo on a new vcpkg install and the newer package versions broke it. Now you can specify dependencies and version requirements in a manifest and use it like a reasonable package manager (ofc, conan has had that long before).
I am taking notes.
really valuable content THANK YOU !
Thanks a lots from vietnam, your videos is so detail and easy to understand
Can you elaborate more on why premake vs CMake? I don't really get the benefit (not an ignorant question, but genuinely no idea).
Also interested
You are doing a great job. I have learnt a lot from your videos. Pls make a series on CONAN package manager.
Thanks a lot🎉
You know the video is about to slap when the thumbnail has The Cherno with that Oppenheimer face.
Yes, I am understanding how to code.
No I had no idea that there was this much variation in how a project can be built....
Looking forward to continuation
I feel like there are more build systems for C++ than there are C++ coding tutorials.
This looks good. However, Can you create a template using CMake? that will be helpful too
Hey there! How can I go about following this but using Visual Studio Code?
Before I used Autotools it was simple to use, but it was limited to Linux build systems, since I expanded my projects to windows and, I migrated the build system to CMake
Could you make a video about how to design a library? or how to set up a library? BTW love your videos, keep going
What's the difference between dist and release?
Dist is what most IDEs and people refer to as Release, I.e. a configuration of full optimizations and no debug information, whereas Release is a Dist build with debug information
@@marcasrealaccount That... doesn't make sense. "Release" should be what you actually, you know, RELEASE into the world, to the public. Adding debug information to release is like... bro, what are you even doing?
You’re less likely to encounter open-source projects using Premake. What about integration with test frameworks?
THANK YOU SO MUCH!!! you're a mind reader 👀
Hey @The Cherno so I followed along your video and used this template. I included GLFW and GLEW into the Core project (as much of the core engine is using these libraries), and I linked the two libraries in the Editor and Runtime projects. The result is a working game engine, but the console is spammed with a bunch of link warning 4217. I'm not one to disable warnings, but the project is running smoothly. After some research I've come to the conclusion that since the Core project (a static library) is including the header files of GLEW and GLFW but not the implementations, when the compiler tries to replace all the function calls from GLEW or GLFW in the Core project files, they're using symbol names (such as "__glewAttachShader") instead of their corresponding function names which get replaced in the Editor and Runtime projects. Is there a work-around to this? Or a way to fix these warnings?
Awesome topics!
Could you make a video on turning cmake libraries to premake and if there are some automated tools.
No offense, but the guy could have used 10 hours of those 2000 hours to learn about design and save himself from all the trouble of duplicating a whole class just to add editor tools to it.
Not even just 1 class. It seems like most classes that *CAN* have editor usefulness are duplicated so they can be used without editor integration. wtf
Damn your so right, but 10h wouldnt be enough time to learn CMake or premake and actuallz apply these core-app concept. I srated this project when i was 15 and learning back then was harder compared to today using ChatGPT (its literally a cheay). I have spent the last week 60hours tlearning CMake and refactoring the whole engine into a core app structure, I LOVE. IT. im so glad i sent him mz project
🥰 Thank U for providing it not only for VS, but others, I personally adore VS code, and most of the time use it!
How come do you go to the end of the line and add a semicolon so fast?
I'm looking everywhere if is there any shortcut to jump to the end of the line. People often suggest Shift+Enter, but it doesn't work.
Omfg i needed this so bad im so bad at being organized like most of us programers that would rather think than organize
I'm new to this now I'm here in this c++ playlist & i know how to make projects with unreal engine, good but still understand nothing from this should i watch the game engine playliist to understand what is in this video, learn get or where to go now.
How would I add sfml to the project template? Where would I add the necessary dll, lib files ect?
This one should be good. I've been waiting a long time to find out what you think is so wrong with CMake :p.
How can I set up that project template manually in Visual Studio?
Am I crazy for preferring VScode and g++? I just set up my build command as a task and whammo.
Why did you choose Premake over CMake?
another really great tool is xmake, which also uses lua but it's kind of its own build system as opposed to a metabuild system
Thoughts on Bazel?
So i was just setting up this template to be used with glfw where i wanted to wrap glfw in the core and only expose a non glfw specific api for the app. One thing i noticed was i had to include the glfw lib as an include of app as well if i included my wrapper even know the wrapper doesn't expose any glfw specific types. Seemingly i should be able to compile glfw into my "core" lib then only include it in the app project right...? Did i set something up wrong or is this not the case?
Since you like premake, have you tried xmake ? Also lua based, but superior in every way.
is it possible to make it work with visual studio code?
Generally a bad idea to teach premake where 90% of all projects are cmake based in the c++ world and the rest are either make or .bat/.sh scripts. I have never seen any major library be built with premake so i would highly recommend investing your time into learning cmake over premake as it really is the industry standard besides its pain points that the cherno has. Bad take honestly on this one
Highly disagree. Cmake is an ancient relic that should be deprecated in any sensible codebase. It‘s not backward compatible, causes many points of breakage, is slow, has terrible documentation and generally causes more issues than it solves.
marry cmake for the shows, have an affair with premake. That what I would like to do.
I do agree with it although premake is lot easier than Cmake in my opinion, also for gamedev it's pretty easy to setup. I just spent 3 days to fully setup a pretty complex HPC based library for my next project in Cmake so I am pretty damn salty because of all the different compilers I have to run and flags I support.
CMake is literally trash. I started with using CMake and it's so confusing, has bad or non existing learning resources and is so complex without any reason. Premake is very beginner friendly and is highly configurable at the same time
I totally agree. If you lack knowledge of any preprocessor or project configuration tool, you are NEVER going to create a complex enough project to need the "simplicity" of Premake over the much faster CMake.
CMake is much more difficult to configure? True, but only for large code projects. For simple or even medium size projects, CMake is the way to go, and one of the key points is that is the industry standard.
DO NOT start with Premake for a simple project, you will spend more time setting all the files and configurations than just running cmake .. && make
Thanks, but CMake, C++20 Modules and vcpkg are doing their job quite well. I'll pass on this one.
What's wrong with using CMake?
I have the same question. Modern CMake is easy (with the correct docs) and straight forward: target_something. Precompiled headers, done. Unity builds, done. Packages, done (conan or Fetch Content). Pre or post build scripts, done.
Finally i can get two the "actually coding" part.
Hi Cherno. Why is Project Red (Cyberpunk ) switching from the RedEngine to Unreal5 ? Why not just improve on their own engine? Isnt making games with your own engine more profitable on the long run? Is it because of some tech like Lumen and Nanite that coders at Project Red can not replicate?
The community support is invaluable. Unreal engine is light years ahead in their development. It just makes sense to use an engine that’s better.
what do you think about cmrk
I have a project were i need to use clion. Premake has no good support for clion. Clion does use cmake which premake can genrate but clion overrides some cmake stuff and i had to make my own version of premake-clion with some ugly hacks. No need to say i switched to cmake in the end.
I used to be on the Premake train. But the integration with IDEs that CMake has just does not compare at all to what Premake offers. Premake is certainly easier for a primarily Visual Studio user for sure. But if I want to develop on Linux and not just "build" on Linux for instance, I want CMake. I'm not some die hard VIM user, I want to be able to use a graphical IDE anywhere.
Cool looking template. What do you usually use to manage dependencies? I recently started using conan for basically all C++ projects, especially when things like OpenGL, ImGUI, etc. are involved.
Vcpkg is another popular dependency management system that works with most IDEs. Also, for mostly simple stuff, git submodule is perfectly fine :)
I think I tried conan before, if I remember right, it has a CMake integration too. I myself use vcpkg (both with CMake, as well as in situations where I am working with visual studio solutions, since it works there too). VCPKG has all the stuff you'd need to OpenGL too, including imgui (with configurable bindings, i.e. imgui[glfw-binding,opengl3-binding] if you want opengl plus glfw setup, which is what I always use). You can also use it as a submodule if you want, since it is itself just compiled with C++, it's easy to set even that up cross-platform in your CMake script so that someone doesn't even have to manually install it.
vcpkg also recently added manifests, which was much needed (i.e., specifying dependencies in a vcpkg.json file in your project, along with version requirements, just like conan does it)
Which vscode color theme been used ?
Can someone tell, why using premake is better than CMake?
For me it seems like almost all libraries are using CMake nowadays.
7:22 fellow chatgpt user :)
What a mess it can be if not structured properly from the beginning. This is when we would have needed a private Cherno-GPT that did what Cherno does here, and more. Well. Cherno-GPT sounds nice.
There is no real “one size fits all” solution for apps. Not everything needs some supporting core library and a whole build system
I'd like to see this project fixed, ESPECIALLY the misspellings :) "darg_and_drop" and "*_initialied" etc are painful to look at!
initia*LIED,* yeah. 😂 No wonder it couldn't find it.
I have a few of those in my own engine. 😳
Would've been nice if Visual Studio was checking some basic grammar (and if you clicked on it, rememberd that you actually WANTED it misspelled). 😩
i fixed already almost everything and implemented this structure!
Say if I have a game engine that has its own renderer and physics (engine) and some other components/engines (audio, animation, etc). Are these components/engines the "Core" and then my Editor or a game can then be the "Apps" and they can include/use the core stuff? Sorry if this is hard to read and understand and my question is that does this follow Core/App architecture? and is that the genreal or recommended approach? If not please explain. Thanks
Hello, its me the creator of this engine haha. I spent the last week 60hours rewriting my whole engine into this core app strucuture, and learning cmake, and as far as i understood yes you are right, it should be implemented this way so you can #include from the static library into thr editor and game, and i LOVE it, no duped files anymore, thanks cherno lol
@@y4ni608 Ok. Thanks a lot for the reply. I’m also trying to do the same thing where I have my renderer as core and then I will try to include it wherever i need (editor, game, etc). But I’m actually just starting out. Wish me luck 😅
@@mehadnadeem4252 Nice, feel free to ask me if youre using CMake i could be helpful to you now haha, good luck :D
Can you do one on cmake too?
So its like inheritance for files? ❤😂
Any integration with linux IDE/Editors? Ive been looking for an alternative to msbuild since ive migrated to linux. Been using CMake and so far it's okay
how do you start adding your own files?
i right-clicked the Core project in the Solution Explorer, and i added one new file in the same directory as Core.cpp and Core.h
its called Logger.h
i included it in Core.cpp like so; #include
DEVELOPING ANYWHERE ELSE this would be fine, but because Visual Studio is beyond annoying, it will not include it and is throwing errors that it cannot find a file name "Logger.h" after pressing F5
i want to throw Visual Studio into a volcano every time i use it
The way I figured out:
1. Right-click the "Core" project -> Choose "Add" -> "New Class..."
2. Then type the name of my class "Foo" or something.
3. Add "Source/Core" to the front of both the Foo.h and Foo.cpp file path input fields. DONT use your file browser to choose the "{Solution}/Core/Source/Core" folder. It does not work. Because... reasons? By default the files just go into the root of that solution, again because... reasons?
4. Click "Ok".
5. Right click "Core" solution again and click "Build".
6. You _should_ then be able to `#include "Core/Foo.h"` in "App.cpp", build it, run it, etc.
I dOn'T kNoW wHy YoU'rE hAvInG tRoUbLe?
\sarcasm incase not obvious.
zig the new cross platform build system for c/c++
a makefile to generate makefiles. ugh.
6:00 outdated
That's nice, I refuse to use modern IDE's still because I'm insane, so I've only really written my own Makefiles so far. Ah the pain. I think I'm going to stop making my own Makefiles now.
Still waiting for that color theme 🙃
Im doing same architecture in Java im creating maven jar projects(core-dist-version(this is also pom)) under parent pom project for version independent projects.
One thing I don't understand in the new stuff, why do they use configuration files that have to be manually typed? Doing that you have to read the documentation. We are in 2023, can't they make a simple GUI to enable us to set everything and not to worry about typing something wrong? (not only premake, even in vscode, many extensions do that)
More stuff to maintain. Text files are just plain text. Everybody can edit it. You can always open and read it
You would still need to read the tool's configuration, also the written config files are easily diffable and manageable through a vcs
Hay not sure if have seen this, but have you had a look at the build chain and tooling that is coming with zig to compile/cross-compile/manage dependencies for C/C++ projects? Think it might make life easier in some way for you.
lmao, up until now i just use python to build my project because im lazy to learn cmake or make
Oof. I wonder what your take on CMake would be.