hanzoai/next.js
2
0
Fork 0
mirror of https://github.com/hanzoai/next.js.git synced 2026-09-24 17:00:51 +00:00
Code Issues Packages Projects Releases Wiki Activity

[turbopack] Switch make_production_chunks to use floats (#95749)

Browse Source 
We've talked about making this switch, making this switch because
without it the diff for https://github.com/vercel/next.js/pull/95157 was
very difficult to read and follow.
This commit is contained in:
Sam Poder
2026-07-13 16:23:33 -07:00
committed by GitHub
parent 63375cd1f7
commit f1185f1341
1 changed files with 26 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
turbopack/crates/turbopack-core/src/chunk/chunking/production.rs
+26 -16
View File
@@ -234,14 +234,15 @@ pub async fn make_production_chunks(
.min(1_000_000) as i64;
// Default `P(N = 1)`: the probability that we request exactly 1 chunk group.
// An integer percentage (0..=100). `firstPageLoadPriority` maps directly to
// it; the default is 67 (~2/3).
let default_p1 = first_page_load_priority.map_or(67, |percent| percent as i64);
// `firstPageLoadPriority` (a config percentage) maps to it; the default is 0.67
// (~2/3).
let default_p1 =
first_page_load_priority.map_or(0.67, |percent| percent as f64 / 100.0);
// `priorityBoost` as an integer percentage to multiply `P(N = 1)` by for priority
// routes; the default is 150 (a 1.5x boost).
let priority_boost_percent =
priority_boost_percent.map_or(150, |percent| percent as i64);
// `priorityBoost` multiplier applied to `P(N = 1)` for priority routes; the default
// is 1.5 (a 1.5x boost).
let priority_boost =
priority_boost_percent.map_or(1.5, |percent| percent as f64 / 100.0);
let mut iterations = 0;
while chunks_to_merge.len() > 1 {
@@ -476,27 +477,36 @@ pub async fn make_production_chunks(
}
let p1 = if is_priority_route {
((default_p1 * priority_boost_percent) / 100).min(100)
(default_p1 * priority_boost).min(1.0)
} else {
default_p1
};
let p2 = 1.0 - p1;
let p2 = 100 - p1;
let c_req = c_req as f64;
let o = o_groups as f64;
let groups = groups as f64;
let rem_g = rem_g as f64;
// pre_dw = PROB_SCALE * d * (rem_g * groups) / o_groups
let pre_dw = p1 * rem_g * c_req
+ p2 * ((o_groups - 1) * c_req
- 2 * (a_rem * a_size + b_rem * b_size));
let d1 = o / groups * c_req;
let d2 = o
* ((o - 1.0) * c_req
- 2.0
* (a_rem as f64 * a_size as f64
+ b_rem as f64 * b_size as f64))
/ (groups * rem_g);
let value = p1 * d1 + p2 * d2;
// It need to have some runtime benefit of merging the chunks
if pre_dw < 0 {
if value < 0.0 {
continue;
}
let value = pre_dw * o_groups / (rem_g * groups);
if let Some((best_i1, best_i2, best_overlap, best_value)) =
best_combination.as_mut()
{
if (overlap.cmp(best_overlap)).then_with(|| value.cmp(best_value))
if (overlap.cmp(best_overlap))
.then_with(|| value.total_cmp(best_value))
== std::cmp::Ordering::Greater
{
*best_i1 = i;