garc0
Will it continue to be updated to C++20?
C++20 supports using string literals in templates and offers more compile-time processing, which can further optimize these codes and processes
Also, I have a question - how to allocate placeholders so memory addresses can be input during runtime?
After reviewing your code implementation, I really like this project and hope it continues to be updated. Regarding placeholders, I was thinking we could use {} for placeholders like std::format does, then input address parameters at runtime, and have the template calculate and write based on the generated offsets - that should work, right?
With C++20 this project will be meaningless, c++20 implement mechanisms for passing string literals in a template, I'm not even talking about compile time functions. It will better to be updated to C++20 because compilers such as clang and msvc have a template layout limit of 1024 or smth You can delete this part with string literals, i added that because of debugging things. About calculating offsets - that what i was thinking about when i wrote this =) Of course i want to implement this feature and this should works, but you need first implement function to access assembled array with your offset I've given up on this project because have a problem with health and radiation, as I get better, I'm increasingly inspired to take on this project
Thank you for your reply, I hope you get better soon.
If possible, could you share the generation tool for the "ctasm_emit.hpp" file?
sure if I can find it, but it was a very ugly nodejs script based on asmjit db
Alright :) While testing short jump instructions, I found the generated offset was wrong. It's a bit hard to read in the template metaprogramming code. No worries, I'm currently slowly breaking down and reading your nested code - it'll take some time.
there a lot of wrong things i wrote, everything was on my mind and nasm 2.14 documentation. Ofc it's hard to read)) But it's a simple project because there a lot of small functions that you can test without debugging, there a simple logic Dunno maybe i'll create a new repo for a new project with C++2b meta To be honest you're the first person that got interested in this)
When we discovered your project, my friend and I were genuinely amazed. If it works as intended, many "external preprocessor assemblers" and "runtime assemblers" could potentially be replaced with native compile-time alternatives, which got us really excited.
Regarding this project, I feel that since C++20, the capabilities of constexpr have improved significantly. Many computations can now be handled by constexpr functions. While they might not be as versatile as template metaprogramming, they allow for more straightforward and readable code, making context handling for labels much more convenient.
While studying your code, I've also attempted to recreate it. After two days, I managed to implement a compile-time lexical parser function. The syntax parsing part is quite complex, and I haven't fully grasped it yet. If you're interested in refactoring, I'd suggest exploring this direction further.
thanks, i love your code, its cool but are you sure you need to use std::variant std::pair and std::string_view here? Is it compile time? Do you use MSVC? I think FixedString will not work there, i heard that is GNU feature
Yes, they are all compile-time features, written and run in MSVC. Some constexpr capabilities may come from C++23 enhancements. In the future, C++ should see more improvements in compile-time function capabilities.
ah it's all about c++23 well ok. If you want to use c++23 you should avoid using templates, compiler lets you do all magic in constexpr function. Ahaha i saw what compiler do with one byte copy. I made an ugly solve to manipulate size of mov that compiler will paste, it was about casting
I tested it and found that changing the compiler standard to C++20 still compiles correctly. C++20 is more powerful than we imagined.
So, would you be interested in collaborating to refactor this project in this direction? Template metaprogramming is quite hard to read, and I'm finding it a bit challenging to recreate it alone...
Sure i want. You want start a new project or just commit to this one?(This project should be c++11-14 supported so std:: libs will not work) Share me your contacts so we can massage more conveniently. Or you can write to me. Which messenger is more convenient? you have telegram?
Let's start a new project. Based on constexpr, reducing the use of template metaprogramming (though it might still be needed in some cases).
Our local laws don't permit using Telegram - could you switch to Discord instead?
Or alternatively, an email address would also work
wait you have no access to telegram with shadowsocks(did you tried - its free like Tor project) but you can login to discord??? In russia opposite anyway discord garco4002
It's not that it won't open—it's that I can't register. I've added you on Discord, please accept. My username is still this one: w4454962.