How do you rewrite C/C++ projects to Rust?

5 days ago

Migrating C/C++ projects to Rust is practical for improving safety and reducing maintenance, but requires a strategic approach.
Successful migrations are incremental, starting with isolated modules integrated into existing build and release processes.
Strong candidates for migration include performance-sensitive, concurrent, or security-critical codebases where maintainability is costly.
Full rewrites are viable only for small codebases with exhaustive tests and stable APIs; otherwise, incremental migration is safer.
Start migration from leaf modules with no dependencies to solve integration challenges early, then expand gradually.
Vertical slice approaches show business value faster but involve complex FFI and less initial safety benefit.
Rust FFI is the most challenging part, requiring careful manual tracking of pointers, ownership, and memory allocation across boundaries.
Prepare by building confidence step-by-step: solve integration problems, keep FFI clear, and ensure team learns Rust.

Code migrations that used to take years can now be completed in weeks or even days using AI agents like Claude Code.
Anthropic successfully migrated large codebases, including Bun's million-line Zig-to-Rust port and a Python-to-TypeScript migration of 165,000 lines.
The core insight is to fix the process (loop) that produces the code, not the code itself.

Bun was rewritten from Zig to Rust in 11 days using AI (Claude's Fable), costing $165,000, which shortened a potential 1-2 year manual migration.
The primary motivation for the rewrite was to eliminate memory-related bugs (e.g., memory leaks, use-after-free) in Zig, as Rust provides memory safety and compiler-enforced guarantees.
The migration process involved: prep work with a porting guide, adversarial reviews, parallel AI agents (64 total), and iterative fixing of compiler errors and test failures.

The transition from frontier model APIs to building your own open-source LLM inference stack often starts due to financial, legal, or engineering concerns.
The initial decision point is whether to buy from managed inference providers or build with open weights and your own hardware, with this post focusing on the build path.
Starting with a single replica using serving engines like vLLM, SGLang, or TensorRT-LLM typically works well, handling batching and KV cache management effectively.

<antirez>6 days ago

Redis Cluster development started in March 2011, but took nearly 4 years to reach its first Release Candidate (3.0.0-rc1) due to early missteps and shifting priorities.
The project was initially rushed because scaling was seen as critical, but Redis itself was too immature, leading to repeated pauses to focus on more fundamental features like persistence, replication, and latency.
The author lacked distributed programming knowledge at the start, resulting in a flawed initial design that was later redesigned after studying the basics.

Current forges like GitHub, GitLab, and Gitea all follow the same pattern set by GitHub, with little connection to actual git usage; most workflows happen inside the forge rather than on the client.
Git is designed for decentralized patch-based kernel development, but most users treat it as a centralized pull/push system, with forges handling identity, reviews, CI/CD, and releases.
Pull request workflows have problems: commits often come in wrong order with fixes piling up; feedback should happen via pre-commit hooks run on the forge before pushing.

Code migrations that used to take years can now be completed in weeks or even days using AI agents like Claude Code.
Anthropic successfully migrated large codebases, including Bun's million-line Zig-to-Rust port and a Python-to-TypeScript migration of 165,000 lines.
The core insight is to fix the process (loop) that produces the code, not the code itself.

Bun was rewritten from Zig to Rust in 11 days using AI (Claude's Fable), costing $165,000, which shortened a potential 1-2 year manual migration.
The primary motivation for the rewrite was to eliminate memory-related bugs (e.g., memory leaks, use-after-free) in Zig, as Rust provides memory safety and compiler-enforced guarantees.
The migration process involved: prep work with a porting guide, adversarial reviews, parallel AI agents (64 total), and iterative fixing of compiler errors and test failures.

The transition from frontier model APIs to building your own open-source LLM inference stack often starts due to financial, legal, or engineering concerns.
The initial decision point is whether to buy from managed inference providers or build with open weights and your own hardware, with this post focusing on the build path.
Starting with a single replica using serving engines like vLLM, SGLang, or TensorRT-LLM typically works well, handling batching and KV cache management effectively.

<antirez>6 days ago

Redis Cluster development started in March 2011, but took nearly 4 years to reach its first Release Candidate (3.0.0-rc1) due to early missteps and shifting priorities.
The project was initially rushed because scaling was seen as critical, but Redis itself was too immature, leading to repeated pauses to focus on more fundamental features like persistence, replication, and latency.
The author lacked distributed programming knowledge at the start, resulting in a flawed initial design that was later redesigned after studying the basics.

Current forges like GitHub, GitLab, and Gitea all follow the same pattern set by GitHub, with little connection to actual git usage; most workflows happen inside the forge rather than on the client.
Git is designed for decentralized patch-based kernel development, but most users treat it as a centralized pull/push system, with forges handling identity, reviews, CI/CD, and releases.
Pull request workflows have problems: commits often come in wrong order with fixes piling up; feedback should happen via pre-commit hooks run on the forge before pushing.

Hasty Briefsbeta