Skip to content
Permalink
master
Switch branches/tags
Go to file
 
 
Cannot retrieve contributors at this time

The build2 Project Roadmap

The aim of this document is to lay out the current build2 project roadmap.

We divide all planned features or areas of work into three categories: "Strategic/Foundational", "Important", and "Nice to Have". Normally, we try to make progress on at least one Strategic/Foundational area in every release, though sometimes we have cleanup releases where the primarily focus is on fixing bugs, paying off accumulated technical debt, etc.

The two common labels you will find next to some features are [WIP] meaning the feature is being actively worked on (and you are welcome to start experimenting with it) as well as [HELP] meaning the feature needs help from the broader community in the form of feedback, expertise (for example, knowledge and experience using a language or tool we are trying to support), and/or actual implementation before it can go forward. Note that this doesn't mean we don't want help on other features, of course. See also Contributing for other ways to contribute to the build2 project.

Strategic/Foundational

  • Conditional dependencies and dependency configuration [WIP]

    Sometimes our projects only need a dependency if certain conditions are met (for example, we are building for Windows). Similarly, sometimes we may need a dependency configured in a certain way (for example, we need a specific feature enabled).

  • Dynamic dependencies in ad hoc recipes and rules [WIP]

    Dynamic dependencies are supported in rules written in C++ but not in ad hoc recipes and rules written in Buildscript. The plan is to extend the Buildscript depdb builtin to support dynamic dependencies, including auto-generated.

  • C++20 modules [WIP] [HELP]

    This is an ongoing effort that is currently blocked by support for modules in the compilers.

    Specifically, for GCC we have complete C++20 modules support sans the ability to handle auto-generated headers. However, the modules support in the compiler itself is not yet in a usable state. We also plan to try to generalize the module mapper with support for auto-generated headers along the P1842 lines.

    For Clang we are waiting for the module mapper which is work in progress.

    Finally, for MSVC we need to understand if it's possible to provide sensible and conforming support without the module mapper (which the MSVC developers have decided not to provide).

    If you would like to help move things forward on any of these three fronts, get in touch.

  • Distributed compilation and caching for C/C++ [WIP]

    We already have a lot of the infrastructure work in place. Specifically, all our compilation goes through partial preprocessing (full preprocessing on MSVC) and we have an accurate idea of all the inputs to each translation unit (headers, modules, compiler options, TU tokens themselves, etc) as well as which of them have changed.

  • Automatic distribution package generation

    We would like to be able to automatically produce binary distribution packages (.deb, .rpm, .msi, etc) from build2 packages and maybe even publish them to cppget.org.

  • Documentation [WIP]

    This includes the documentation for the missing pieces of the toolchain itself (mostly the build system) as well as the documentation extraction (from source code) that everyone can use to document their build system modules (and maybe C++ projects in general).

Important

  • Assembler support [HELP]

    We need first-class assembler support in the bin/cc modules. See issue #162 for background.

  • Autoconf emulation module [HELP]

    We have started work on libbuild2-autoconf, a build system module that provides a rule for processing GNU Autoconf config.h.in files as well as their CMake and Meson flavors. We need help populating it with built-in checks.

  • Lockfile

    We would like the package manager to provide the lockfile functionality.

  • Config initializer

    We would like to have a higher-level (than compiler options) mechanism for specifying the desired configuration.

    See issue #11.

  • Support for static analysis [HELP]

    We would like to have a general approach for running static analysis tools.

    See issue #30 for some background as well as to contribute usage experience, ideas, etc.

  • Incremental testing

    We would like to be able to run tests only for targets that have changed. This will rely heavily on our existing work on Testscript, specifically, the fact that the test inputs/outputs are well-encapsulated.

Nice to Have

See the Issues List for some of the nice-to-have features.