[Umbrella] Implement PendingUpdate subclasses
- 主要语言
- C++
- 星标
- 221
- 派生
- 124
- 平均合并
- 1 天 17 小时
- 30 天内合并 PR
- 21
描述
This issue tracks the progress of implementation of `PendingUpdate` subclasses. To implement each subclass, we need to make sure that following goals are met:
1. Implement respective functions in the [`TableMetadataBuilder`](https://github.com/apache/iceberg-cpp/blob/dbcbdf23ff7ebe5510431715fd66c7841df3d9fd/src/iceberg/table_metadata.h#L196) for this kind of update.
2. Implement respective [`TableUpdate`](https://github.com/apache/iceberg-cpp/blob/main/src/iceberg/table_update.h) subclass (if exists) to apply uncommitted changes to TableMetadataBuilder and generate update requirements.
3. Expose the new update as an interface to [`Table`](https://github.com/apache/iceberg-cpp/blob/main/src/iceberg/table.h) and [`Transaction`](https://github.com/apache/iceberg-cpp/blob/main/src/iceberg/transaction.h) classes.
Below is the list of subclasses that are expected to add. They should be put into the `iceberg/update` subdirectory.
### Medadata Update:
Unlike Java, we don't need pure interface for these classes.
- [x] UpdateProperties @WZhuo
- [x] ReplaceSortOrder @HeartLinked
- [x] UpdateSchema @gty404
- [x] UpdatePartitionSpec @zhjwpku
- [x] ManageSnapshots (<- SnapshotManager) @zhjwpku
- [x] SetSnapshot @HeartLinked
- [x] UpdateLocation (<- SetLocation) @HeartLinked
- [x] UpdatePartitionStatistics @HeartLinked
- [x] UpdateSnapshotReference @zhjwpku
- [x] UpdateStatistics @HeartLinked
- [x] ExpireSnapshots @dongxiao1198 @shangxinli
### Data Update:
For these classes, we still need intermediate interfaces and classes to consolidate common logic.
- [x] SnapshotUpdate(SnapshotProducer) @zhjwpku
- [x] FastAppend @zhjwpku
- [ ] MergingSnapshotProducer @gty404 (planned in 0.3.0)
- [ ] OverwriteFiles @gty404 (planned in 0.3.0)
- [ ] RowDelta
- [ ] MergeAppend
- [ ] DeleteFiles
- [ ] RewriteFiles
- [ ] ReplacePartitions
- [ ] RewriteManifests
- [ ] CherryPickOperation
贡献指南
这个仓库没有索引到贡献指南
调研方向
从 iceberg/update 目录中列出的未勾选子类开始,并将它们与已完成的实现进行比较。阅读 src/iceberg/table_metadata.h、src/iceberg/table_update.h、src/iceberg/table.h 和 src/iceberg/transaction.h,以跟踪所需的元数据、更新和公共接口。选定的子类已实现,并通过相关的 Table 和 Transaction API 对外提供,即表示完成。
由索引模型根据 Issue 内容生成。
评估
- 技术栈
- cpp
- 领域
- databases
- Issue 类型
- 功能
- 难度
- 5/5
- 预计耗时
- 一周以上
- 活跃度
- 冷清
- 描述清晰度
- 基本清楚
- 新手友好度
- 35/100