Installer libgrape-lite avec Homebrew

Liens du projet

• Page d'accueil
• Dépôt

Commandes et alias

• run_app

Historique du projet

The repository README ties libgrape-lite to GRAPE, the Graph Parallel Processing Engine described in SIGMOD and VLDB work from 2017. The project presents itself as a lightweight, open-source implementation of the GRAPE core rather than a full graph platform.

GraphScope documentation says its analytical engine originated from GRAPE, and that a lightweight version of GRAPE is open-sourced as libgrape-lite. GraphScope extends it with mutable fragments, vineyard support, and service-mode features.

Historique d'adoption

libgrape-lite's adoption story is mostly inside Alibaba and GraphScope rather than broad end-user CLI use. The README's badge says it enables GraphScope, and GraphScope's glossary calls libgrape-lite a dependency that serves as the analytical engine core.

The v0.1.0 tag points to a 2021-10-29 commit, and subsequent tags through v0.3.5 show a small versioned release series. Homebrew packaging makes the C++/MPI graph library available outside the GraphScope source tree for users who want the lighter standalone component.

Modes d'utilisation

Users build libgrape-lite with CMake, a C++11 compiler, MPI, and glog, then run example analytical applications such as single-source shortest path, weakly connected components, PageRank, local clustering coefficient, label propagation, and breadth-first search.

The project also includes LDBC benchmark support, optional GPU graph processing with CUDA and NCCL, and a GNN sampler example, so the package is most relevant to graph analytics researchers and infrastructure developers.

Pourquoi les passionnés de paquets s'y intéressent

The package is niche but interesting because it is a standalone slice of a larger graph-computing platform. Packaging libgrape-lite separately lets users experiment with GRAPE-style graph analytics without installing all of GraphScope.

Its dependency profile is very package-manager-shaped: C++11, MPI, glog, optional jemalloc, optional CUDA/NCCL, and benchmark/examples. That makes it a small formula with heavyweight ecosystem implications.

Chronologie

• 2017: GRAPE publications appeared at SIGMOD and VLDB, with the README citing Best Paper and Best Demo awards.
• 2021-10-29: The v0.1.0 tag points to a commit from this date.
• 2021: GraphScope was presented in VLDB and open sourced, with its analytical engine tied to GRAPE.
• 2026-07-01: The batch input listed Homebrew packaging for standalone libgrape-lite.

Related projects

• GraphScope is the larger graph-computing system whose analytical engine builds on GRAPE/libgrape-lite.
• Vineyard is named by GraphScope documentation as an extension point around the analytical engine.
• LDBC Graphalytics, CUDA, NCCL, MPICH, OpenMPI, glog, and GraphScope's GRAPE-JDK appear in the project's documented ecosystem.

Sources

• https://alibaba.github.io/libgrape-lite
• https://github.com/alibaba/GraphScope/blob/main/analytical_engine/README.md
• https://github.com/alibaba/GraphScope/blob/main/docs/overview/glossary.md
• https://github.com/alibaba/libgrape-lite
• https://github.com/alibaba/libgrape-lite/blob/master/README.md