docs: add docs for the merge_mode option

docs/nightly/en/reference/sql/compatibility.md

@@ -9,6 +9,7 @@ GreptimeDB supports a subset of ANSI SQL and has some unique extensions. Some ma
 2. Insert data: Consistent with ANSI SQL syntax, but requires the `TIME INDEX` column value (or default value) to be provided.
 3. Update data: Does not support `UPDATE` syntax, but if the primary key and `TIME INDEX` corresponding column values are the same during `INSERT`, subsequent inserted rows will overwrite previously written rows, effectively achieving an update.
    * Since 0.8, GreptimeDB supports [append mode](/reference/sql/create#create-an-append-only-table) that creates an append-only table with `append_mode="true"` option which keeps duplicate rows.
+   * GreptimeDB supports [merge mode](/reference/sql/create#create-an-append-only-table) that creates a table with `merge_mode="last_non_null"` option which allow updating a field partially.
 4. Query data: Query syntax is compatible with ANSI SQL, with some functional differences and omissions.
    * Does not support views.
    * TQL syntax extension: Supports executing PromQL in SQL via TQL subcommands. Please refer to the [TQL](./tql.md) section for details.

docs/nightly/en/reference/sql/create.md

@@ -98,7 +98,8 @@ Users can add table options by using `WITH`. The valid options contain the follo
 | `compaction.twcs.max_inactive_window_files` | Max num of files that can be kept in inactive time window.         | String value, such as '1'. Only available when `compaction.type` is `twcs`. |
 | `compaction.twcs.time_window` | Compaction time window    | String value, such as '1d' for 1 day. The table usually partitions rows into different time windows by their timestamps. Only available when `compaction.type` is `twcs`.  |
 | `memtable.type` | Type of the memtable.         | String value, supports `time_series`, `partition_tree`. |
-| `append_mode`           | Whether the table is append-only     | String value. Default is 'false', which removes duplicate rows by primary keys and timestamps. Setting it to 'true' to enable append mode and create an append-only table which keeps duplicate rows.     |
+| `append_mode`           | Whether the table is append-only     | String value. Default is 'false', which removes duplicate rows by primary keys and timestamps according to the `merge_mode`. Setting it to 'true' to enable append mode and create an append-only table which keeps duplicate rows.     |
+| `merge_mode`           | The strategy to merge duplicate rows     | String value. Only available when `append_mode` is 'false'. Default is `last_row`, which keeps the last row for the same primary key and timestamp. Setting it to `last_non_null` to keep the last non-null field for the same primary key and timestamp.     |
 | `comment`           | Table level comment   | String value.      |
 
 #### Create a table with TTL
@@ -149,6 +150,67 @@ CREATE TABLE IF NOT EXISTS temperatures(
 ) engine=mito with('append_mode'='true');
 ```
 
+#### Create a table with merge mode
+Create a table with `last_row` merge mode, which is the default merge mode.
+```sql
+create table if not exists metrics(
+    host string,
+    ts timestamp,
+    cpu double,
+    memory double,
+    TIME INDEX (ts),
+    PRIMARY KEY(host)
+)
+engine=mito
+with('merge_mode'='last_row');
+```
+
+Under `last_row` mode, the table merges rows with the same primary key and timestamp by only keeping the latest row.
+```sql
+INSERT INTO metrics VALUES ('host1', 0, 0, NULL), ('host2', 1, NULL, 1);
+INSERT INTO metrics VALUES ('host1', 0, NULL, 10), ('host2', 1, 11, NULL);
+
+SELECT * from metrics ORDER BY host, ts;
+
++-------+-------------------------+------+--------+
+| host  | ts                      | cpu  | memory |
++-------+-------------------------+------+--------+
+| host1 | 1970-01-01T00:00:00     |      | 10.0   |
+| host2 | 1970-01-01T00:00:00.001 | 11.0 |        |
++-------+-------------------------+------+--------+
+```
+
+
+Create a table with `last_non_null` merge mode.
+```sql
+create table if not exists metrics(
+    host string,
+    ts timestamp,
+    cpu double,
+    memory double,
+    TIME INDEX (ts),
+    PRIMARY KEY(host)
+)
+engine=mito
+with('merge_mode'='last_non_null');
+```
+
+Under `last_non_null` mode, the table merges rows with the same primary key and timestamp by keeping the latest value of each field.
+```sql
+INSERT INTO metrics VALUES ('host1', 0, 0, NULL), ('host2', 1, NULL, 1);
+INSERT INTO metrics VALUES ('host1', 0, NULL, 10), ('host2', 1, 11, NULL);
+
+SELECT * from metrics ORDER BY host, ts;
+
++-------+-------------------------+------+--------+
+| host  | ts                      | cpu  | memory |
++-------+-------------------------+------+--------+
+| host1 | 1970-01-01T00:00:00     | 0.0  | 10.0   |
+| host2 | 1970-01-01T00:00:00.001 | 11.0 | 1.0    |
++-------+-------------------------+------+--------+
+```
+
+
 ### Column options
 
 GreptimeDB supports the following column options:

docs/nightly/zh/reference/sql/compatibility.md

@@ -9,6 +9,7 @@ GreptimeDB 支持的 SQL 是 ANSI SQL 的子集，并且拥有一些特有的扩
 2. 插入新数据： 与 ANSI SQL 语法一致，但是强制要求提供 `TIME INDEX` 列值（或默认值）。
 3. 更新：不支持 `UPDATE` 语法，但是在 `INSERT` 的时候，如果主键和 `TIME INDEX` 对应的列值一样，那么后续插入的行将覆盖以前写入的行，从而变相实现更新。
     * 从 0.8 开始， GreptimeDB 支持 [append 模式](/reference/sql/create#创建-Append-Only-表)，创建时指定`append_mode = "true"` 选项的表将保留重复的数据行。 
+    * GreptimeDB 支持 [merge 模式](/reference/sql/create#create-an-append-only-table)，该模式使用 `merge_mode="last_non_null"` 选项创建表，允许部分更新字段。
 4. 查询：查询语法兼容 ANSI SQL，存在部分功能差异和缺失
     * 不支持视图
     * TQL 语法扩展：TQL 子命令支持在 SQL 中执行 PromQL，详细请参考 [TQL](./tql.md) 一节。

docs/nightly/zh/reference/sql/create.md

@@ -100,7 +100,8 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name
 | `compaction.twcs.max_inactive_window_files` | 非活跃时间窗口内的最大文件数         | 字符串值，如 '1'。只在 `compaction.type` 为 `twcs` 时可用 |
 | `compaction.twcs.time_window` | Compaction 时间窗口    | 字符串值，如 '1d' 表示 1 天。该表会根据时间戳将数据分区到不同的时间窗口中。只在 `compaction.type` 为 `twcs` 时可用 |
 | `memtable.type` | memtable 的类型         | 字符串值，支持 `time_series`，`partition_tree` |
-| `append_mode`           | 该表是否时 append-only 的     | 字符串值. 默认为 'false'，表示会根据主键和时间戳对数据去重。设置为 'true' 可以开启 append 模式和创建 append-only 表，保留所有重复的行     |
+| `append_mode`           | 该表是否时 append-only 的     | 字符串值。默认值为 'false'，根据 'merge_mode' 按主键和时间戳删除重复行。设置为 'true' 可以开启 append 模式和创建 append-only 表，保留所有重复的行    |
+| `merge_mode` | 合并重复行的策略 | 字符串值。只有当 `append_mode` 为 'false' 时可用。默认值为 `last_row`，保留相同主键和时间戳的最后一行。设置为 `last_non_null` 则保留相同主键和时间戳的最后一个非空字段。 |
 | `comment`           | 表级注释     | 字符串值.      |
 
 #### 创建指定 TTL 的表
@@ -151,6 +152,70 @@ CREATE TABLE IF NOT EXISTS temperatures(
 ) engine=mito with('append_mode'='true');
 ```
 
+#### 创建带有 merge 模式的表
+
+创建一个带有 `last_row` merge 模式的表，这是默认的 merge 模式。
+
+```sql
+create table if not exists metrics(
+  host string,
+  ts timestamp,
+  cpu double,
+  memory double,
+  TIME INDEX (ts),
+  PRIMARY KEY(host)
+)
+engine=mito
+with('merge_mode'='last_row');
+```
+
+在 `last_row` 模式下，表会通过保留最新的行来合并具有相同主键和时间戳的行。
+
+```sql
+INSERT INTO metrics VALUES ('host1', 0, 0, NULL), ('host2', 1, NULL, 1);
+INSERT INTO metrics VALUES ('host1', 0, NULL, 10), ('host2', 1, 11, NULL);
+
+SELECT * from metrics ORDER BY host, ts;
+
++-------+-------------------------+------+--------+
+| host  | ts                      | cpu  | memory |
++-------+-------------------------+------+--------+
+| host1 | 1970-01-01T00:00:00     |      | 10.0   |
+| host2 | 1970-01-01T00:00:00.001 | 11.0 |        |
++-------+-------------------------+------+--------+
+```
+
+
+创建带有 `last_non_null` merge 模式的表。
+
+```sql
+create table if not exists metrics(
+  host string,
+  ts timestamp,
+  cpu double,
+  memory double,
+  TIME INDEX (ts),
+  PRIMARY KEY(host)
+)
+engine=mito
+with('merge_mode'='last_non_null');
+```
+
+在 `last_non_null` 模式下，表会通过保留每个字段的最新值来合并具有相同主键和时间戳的行。
+
+```sql
+INSERT INTO metrics VALUES ('host1', 0, 0, NULL), ('host2', 1, NULL, 1);
+INSERT INTO metrics VALUES ('host1', 0, NULL, 10), ('host2', 1, 11, NULL);
+
+SELECT * from metrics ORDER BY host, ts;
+
++-------+-------------------------+------+--------+
+| host  | ts                      | cpu  | memory |
++-------+-------------------------+------+--------+
+| host1 | 1970-01-01T00:00:00     | 0.0  | 10.0   |
+| host2 | 1970-01-01T00:00:00.001 | 11.0 | 1.0    |
++-------+-------------------------+------+--------+
+```
 
 ### 列选项