Difference between revisions of "Cpp features blocked by version"

From The Battle for Wesnoth Wiki
(Features: Add some notes)
(Features)
Line 31: Line 31:
 
|-
 
|-
 
| Standard feature check macros || Ability to enable features on an individual basis without doing a whole-language flag || Unknown || Unknown || Unknown || Clang and likely GCC already support them on older versions as an extension.
 
| Standard feature check macros || Ability to enable features on an individual basis without doing a whole-language flag || Unknown || Unknown || Unknown || Clang and likely GCC already support them on older versions as an extension.
 +
|-
 +
| std::filesystem || Ability to enable features on an individual basis without doing a whole-language flag || Unknown || Just need to target 10.15 || Unknown || We do have boost....
 
|}
 
|}

Revision as of 23:08, 8 August 2024

Summary

This is a listing of C++ features that would be useful to have but can't yet be used. The Feature column is the language or standard library feature (https://en.cppreference.com/w/cpp/compiler_support is a good reference). The Benefit column is what having this feature will allow to be done. The remaining three columns are for the minimum requirements for the feature to be available on each platform. The minimum requirements can be anything - Visual Studio version, compiler version, OS version, etc.

Benefits that don't justify version increases

This is a non-exhaustive list:

  • Improving compile speed.
  • Pretty much any standard library feature that we can just as well get through boost or other means.
  • Core language features that allow writing more succinct code.

Benefits that do justify version increases

This is a non-exhaustive list:

  • Required to fix a bug that has a significant impact to players.
  • Core language features that make classes of bugs impossible or significantly harder to accidentally introduce.
  • Required to fix a relevant security vulnerability.
  • A group of multiple smaller features that individually don't justify a version increase, if there's minimal to no impact to developers and players.

Features

Feature Benefit Windows macOS Linux Notes
Modules Faster compile time Unknown Unknown Unknown While part of C++20, usable support seems to be progressing very slowly.
std::starts_with and ends_with Simplies string checking in various places, including filesystem MSVC STL 19.21 (VS 2019 16.1) Apple Clang 10.0.0 (Xcode 10.0 beta (10L176w), 10.0 (10A255), 10.1 (10B61)) GCC libstdc++ 9/Clang libc++ 6 Boost offers similar functions.
std::format Cleans up our numerous uses of manual string concatenation, especially with pango strings in the UI Unknown Unknown Unknown The wesnoth variable substitution facility offers similar functions, though with a little less versatility (but it could be extended).
std::span Potential to reduce templates where we only need a container view Unknown Unknown Unknown
Concepts Cleaner template requirements (increased code clarity) Unknown Unknown Unknown
Standard feature check macros Ability to enable features on an individual basis without doing a whole-language flag Unknown Unknown Unknown Clang and likely GCC already support them on older versions as an extension.
std::filesystem Ability to enable features on an individual basis without doing a whole-language flag Unknown Just need to target 10.15 Unknown We do have boost....