World's first open-source enterprise world model.
Rust agent-memorybitemporalgraphragknowledge-baseknowledge-graphllmontologypgvectorpostgresqlragrustself-hostedsemantic-searchtemporal-knowledge-graphworld-model
Run CI on dev pushes and save the Rust cache only there so PRs can finally restore it (#873)
WaylandYang8c940a4a99d6
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.ymlindex 62cb653b..06d3ec17 100644--- a/.github/workflows/ci.yml+++ b/.github/workflows/ci.yml@@ -2,7 +2,11 @@ name: CIon:push:- branches: [main]+ # dev 也要跑,为的是缓存:GitHub 只让 PR 读到自己、base 分支和默认分支的缓存,+ # 而此前没有任何一次 CI 在 dev 上跑过,于是 PR 之间互相看不见对方存的东西,+ # 每次都从零编 300 多个依赖——今天 12 次 backend 运行,恢复缓存全是 No cache found。+ # 顺带把「合并后的 dev 从没被 CI 跑过」这个口子堵上(见下面 migrations job 的注释)+ branches: [main, dev]pull_request:# **显式只读。** 没有这一块时它继承仓库级默认,而那个默认要为@@ -38,7 +42,12 @@ jobs:- uses: dtolnay/rust-toolchain@stablewith:components: rustfmt, clippy+ # 只有 push(dev / main)那次运行才写缓存,PR 只读。此前每个 PR 各存约 2GB,+ # 仓库上限 10GB,五六个 PR 就挤满,按 LRU 淘汰——同一个 PR 第二次跑时自己上次+ # 存的也已经没了:17 条缓存的最后访问时间全等于创建时间,从来没人读到过- uses: Swatinem/rust-cache@v2+ with:+ save-if: ${{ github.event_name == 'push' }}- name: Install PDF text extraction toolsrun: sudo apt-get update && sudo apt-get install -y poppler-utils poppler-data- name: Format@@ -93,7 +102,10 @@ jobs:exit 1fi- uses: dtolnay/rust-toolchain@stable+ # 只在 push 时写缓存,理由见 backend job 同一处的注释- uses: Swatinem/rust-cache@v2+ with:+ save-if: ${{ github.event_name == 'push' }}- name: 装 sqlx-clirun: cargo install sqlx-cli --no-default-features --features rustls,postgres --locked- name: 全新库上跑一遍