How to build a simple C++ project with CMake? What about advanced projects? #programming #tech #softwaredevelopment #cmake #cpp Codes are no longer available. SECOND EDITION IS OUT! Get it here: packt.link/G56Ny
As software grows in capabilities it grows in complexity and build setups must deal with it. However, if you compress complexity at one end you will make room for more complexity at another end. The cycle is endless, there is no escape from the software samsara. But... for those of us who still want computers, gotta make do with 400 page manuals. That seems like a decent investment in front of endless hours of bestemmie one can get into by using other tools.
I'm doing undergrad research and I need to write c++ code with more industrial standards. this video was SO helpful, i feel like im going in the right direction. Thanks!
Great video, thanks! I have ordered your book and have added it to my companies recommended reading list, hopefully this will drive some additional orders!
Really great book, started learning about CMake and stumbled upon this book recommended from a presentation I went to about it, and it's an amazing book and about so much more than CMake and a whole view of how building things in C++ work and what's involved in it, I'm definitely recommending this to my colleagues because it's a good resource for them to get a better view of everything involved in compilation/building and the pitfalls/issues and how to tune it alongside just learning proper CMake Ps: Any option to get non-expensive shipping to Belgium? I'd love to get a copy at my office for my colleagues but the shipping is like more than half the cost
I'm glad you're enjoying the book! I don't know what book shops have their presence in Belgium, but some of the biggest ones have license to sell books from Packt in their countries. See if you can find any. Another solution might be buying on Belgian Amazon, which works great if you have a friend with Prime who could order the book for you. www.amazon.com.be/dp/1801070059 If all fails - maybe an ebook would be more suitable? Good luck with your C++!
@@SmokCode That works! Amazon had it too since they're in Belgium now. I just had one more question I didn't see addressed in your book and I'm having difficulties finding an answer for. How do you deal with targets with hundreds of source files you want to logically split up a bit as some components/folders might get extracted later on? A giant CMakeLists seems a bit icky, object libraries didn't seem to be a great solution either because then I need to link my interface library to every single one of those. So I don't really know if there's any other good constructs to make a sort of division within a single target
Within a single target: you can define variables with filenames and add those to sources list. I'd recommend going with libraries though. If there is a logical relationship between files, it creates an opportunity to test them together in integration tests and writing them for smaller scopes is more manageable and quicker to codeloop.
There is only one criticism I have of the book: spacing. The book has a crammed look, I feel it would be more attractive if there were more empty lines. I'm referring to the epub version as read by Sumatra.
I use an sh-file (and on Windows a bat) that has the command to compile the project, including linking to libs. To further simplify compilation I have a single compilation unit that includes all the other c- or cpp-files.
I think you might be misusing the compiler a bit. If you're really including other CPP files, you're treating multiple compilation units as one. This means that a lot of code is being recompiled when it needn't. This approach may work for small projects, but as your codebase grows it slows down exponentially. Consider reading up on SOLID principles, you might be in for a treat!
@@SmokCode I was being general. I work in game development, and depending on project size I either have two compilation units, system and game, or four, system, graphics, audio and game. Here, system is an API layer I've written that connects the game code to the OS and hardware, and this almost never changes. Graphics, audio and game change often, but in small projects I put them into the same compilation unit. In small projects it takes less than a second to compile and link, and in larger projects so far it's always been less than 10 seconds. But even if I did full recompilations, it seems that the linker still takes almost half of the compilation time. SOLID consist of five principles, but I assume you're particularly referring to dependency inversion, since we're talking about project structuring? In game development it's mostly about knowing the hardware you're developing for, and knowing your data. In the cases where I need this kind of decoupling that DIP offers, I can just use a function pointer. Well, I guess that's a form of dependency inversion too. In general, I don't want to use principles that sacrifice runtime efficiency for a benefit that I can get in other ways. EDIT: I forgot to write that many projects I've seen are structured around an object oriented model, and there's a compilation unit for each different class, and there are dozens, maybe even hundreds, or in large projects thousands of classes. I realize that having only one compilation unit per entire subsystem is very different and might seem strange :)
There are plenty of tutorials on cmake and I've scanned quite a few before this one. You answered the most important question within the first minute, what the fuck is a cmake anyways? Thank you
I love that no one really know what cmake does and therefore pretend to explain it but actually gloss over it by saying "cmake will create the appropriate build system" instead.
ive ordered this book a 2-3 weeks ago on amazon yet it has unfortunately not come yet, even admitting its late... dp you know why this is? im not here to rant id just lilke to really know
It’s good that someone writes updated book but - Why would in 2023 I use archaic build system with autistic syntax language? What if we use xmake? Something based on real language and not as slow as Scons ?
Some of chapters will. But not all of the book, since I focus on practical use which often means employment of external tools which are often language specific. See the list of chapters and assume that first part works for C, second somewhat works, and third mostly doesn't. If that's enough for you, you'll have a good use of the book. Thanks for your question.
Kupiłem bo potrzebowałem zacząć z tym tematem. Zdecydowanie lektura nie dla początkujących, po przeczytaniu pierwszych 100 stron nie ma żadnego progresu. Wydaje się że najlepiej najpierw przejrzeć nawet przestarzałe przykłady z sieci, filmik jak hindus stawia VSC z Clang, cmake i ninja, albo inny stack, żeby zobaczyć efekty jak to w ogóle działa i dalej książkę ale najlepiej od środka. No trudno narazie kasa w błoto, może kiedyś temat dojrzeje i będą efekty. Wracam do hinduskich filmów.
Trochę taki feedback bez konkretu. Nie wiadomo czego dokładnie zabrakło, ani jaki był cel. Książka ma przecież image dockera z pełnym stackiem i można wypróbować wszystko na gotowym.
@@SmokCode I find it far simpler to say something like "gcc hello.c ..." or "cl hello.c ...". It's not much work. It's usually a line or two. It doesn't require me to then install, maintain, and learn an entire suite of commands and syntax. In other words, things like cmake feel redundant. They are solving a problem, rather poorly, when it has already been solved and isn't very complicated to start with.
Every time you type a command you're redoing the same work you did previously. You may introduce mistakes, and it's not viable to do it on large scale. Even hobby projects often have hundreds of files, and commercial solutions have orders of magnitude more. Keeping track of lengthy commands that change over time is difficult and recompiling everything by default wastes time that quickly adds up.
@@SmokCode Yeah imagine typing "gcc ...." to compile a huge and complex project lol. Maybe he is basing his comments on the project you demonstrated on the video, but this is not even close to real scenarios where you need to manage compilations to different targets and manage dependencies for example! Will be nice if you show some real (open source) projects and how they handle their compilation process. I think serenityos uses cmake to generate the build tree, I don't know if you know this project though.
Perhaps I'm suffering from misophonia but the music in the background is REALLY annoying. I'm only 3 minutes in and I don't know if I will be able to get to the end.
