kzc

This proposed `void 0` transform could work if it were applied after the final compress pass but before mangle. I verified that it produces a detectable post-gzip improvement in some...

@Skalman Sure, go for it. It would be a safe transform but should be gated by some new `compress` option so that it can disabled if need be. I made...

Look at `make_sym` in `lib/compress.js` as an example of how to create a local variable in an already existing `var` statement.

The biggest obstacle to implementing this undefined transform is to not spend excessive CPU scanning each and every function for `undefined` and/or `void 0` use. If it is too expensive...

> It looks like AST_Lambda is the base for all functions, so hopefully it would work automatically if I just check this. If you're looking for an existing `AST_Var` that's...

> This transform only makes sense when mangle is enabled, so I suppose the default enabled-ness value should depend on mangle. Just treat it like a normal `compress` option ignoring...

The webkit VM used by bun appears to be slower than the V8 engine for the test above when creating a large number of transient substrings. I don't know whether...

private branch or something? ``` 404 - page not found Cannot find a valid ref in 7a0b13626e5db69aa5a32d037431d381df5dfb61/Source/JavaScriptCore/jit/JITOperations.cpp ```