|
|
|
|
|
|
by iainmerrick
1617 days ago
|
|
|
you get a compiler capable of compiling Zig code stupidly fast, and that's even without factoring in incremental compilation with in-place binary patching, with which we're aiming for sub-millisecond rebuilds of arbitrarily large projects That sounds great! But at the same time people in other threads here are talking about 1-3 second compilation times for Advent of Code solutions (which I presume are smallish). Can you summarise where that really fast compiler comes from, to save me searching through that talk video? Is this something that everyday users will be able to use in typical workflows? |
|
|
https://kristoff.it/blog/zig-new-relationship-llvm/
Long story short, we're currently working on a self-hosted implementation of the compiler and what people are using now is the old C++ implementation. As soon as the new compiler is feature-complete enough, we'll start shipping it and we expect much better compilation speeds, which will be even greater speed for debug builds once the native (i.e., non-llvm) backends catch up as well.
Latest progress update on this work: https://twitter.com/andy_kelley/status/1481862781380874240?s...