目前，timeline.ColumnMap的定义是:

type ColumnMap map[string]interface{}

运行期，会对value做check。很容易出现类型不匹配导致 MissInUse 错误。

建议，采用struct，定义强类型:

type ColumnMap struct {
	int64Fields   map[string]int64
	bytesFields   map[string][]byte
	stringFields  map[string]string
	float64Fields map[string]float64
}

并辅助一些setter，简化操作:

func (cm *ColumnMap) AddInt64Column(key string, val int64) {
	cm.int64Fields[key] = val
}

func (cm *ColumnMap) AddStringColumn(key string, val string) {
	cm.stringFields[key] = val
}

map[X-Ots-Requestid:[0005698d-9dce-b347-0cc3-980a09c33ec8] X-Ots-Contenttype:[protocol buffer] X-Ots-Date:[2018-04-11T07:30:42.430654Z] Date:[Wed, 11 Apr 2018 07:30:42 GMT] Connection:[keep-alive] Authorization:[OTS *************] X-Ots-Contentmd5:[TzgBb4fK4cxD8Ldjw4JGHQ==]]

显示通过x-ots-requestid 拿数据失败

The dependency satori/go.uuid is no longer maintained satori/go.uuid#84 and it's current pinning on v1.2.0 suffers from this vulnerability satori/go.uuid#73.

Suggested fixes include either pinning the latest commit on the master branch of satori/go.uuid where this issue has been fixed satori/go.uuid@75cca53 or move to a maintained replacement such as gofrs/uuid or google/uuid, which also resolves a number of other bugs.

The only change besides changing the pinning would be this line:

--- tunnel/api.go.orig	2021-03-02 16:37:50.890286864 +1100
+++ tunnel/api.go	2021-03-02 16:38:28.083334608 +1100
@@ -163,7 +163,7 @@ func (api *TunnelApi) doRequestInternal(
 		hreq.Header.Set(xOtsHeaderStsToken, api.securityToken)
 		otshead.set(xOtsHeaderStsToken, api.securityToken)
 	}
-	traceId := uuid.NewV4()
+	traceId := uuid.Must(uuid.NewV4())
 	hreq.Header.Set(xOtsHeaderTraceID, traceId.String())
 	otshead.set(xOtsHeaderTraceID, traceId.String())

goroutine 1 [running]:
github.com/aliyun/aliyun-tablestore-go-sdk/tablestore.(*TableStoreClient).DescribeTable(0xc420396150, 0xc42013fbb8, 0x0, 0x0, 0x2)
/Users/redchen/go/src/github.com/aliyun/aliyun-tablestore-go-sdk/tablestore/api.go:367 +0x461
ots.DescTable(0xb84cc0, 0xc4201eaa40, 0x6, 0x2, 0x8f2b90, 0x15)
/Users/redchen/workspace/go/aliyun-tablestore-cli/src/ots/ots.go:88 +0x110
commands.glob..func11(0xc4202fa1c0)

• go版本: 1.17
• sdk版本: v5.0.6
• 代码如下

batchWriteReq := &tablestore.BatchWriteRowRequest{}

for _, data := range datas {
      putRowChange := new(tablestore.PutRowChange)
      putRowChange.TableName = table_im_user_pool
      putPk := new(tablestore.PrimaryKey)
      putPk.AddPrimaryKeyColumn("pk1", "xxxx")
      putPk.AddPrimaryKeyColumnWithAutoIncrement("pk2")
      putRowChange.PrimaryKey = putPk
      putRowChange.AddColumn("col", "xxxxx")
      putRowChange.SetCondition(tablestore.RowExistenceExpectation_IGNORE)
      // 这里设置了
      putRowChange.SetReturnPk()
      batchWriteReq.AddRowChange(putRowChange)
}

// resp内并没有返回主键
resp, err := client.BatchWriteRow(batchWriteReq)

请问有计划支持并发导出数据吗？

我在参考这个实现rust，发现这个函数的逻辑实现不完整．

１．　根据plainbuffer的定义，https://help.aliyun.com/document_detail/50600.html?spm=a2c4g.11186623.6.828.d1c27c55Bc5la6　　PK也可能不给，这里一定写出去了TAG_PK
2. 没有处理tag_delete_marker

SDK should export error value stead of let users using string compare err.Error()

现在项目没有做格式转化，导致代码本自动format之后产生较多的改动

Two primary keys: ProdId string, CommentId int64 (NOT incremental)

When query multiple rows by ProdId, with Min and Max of CommentId, panic when kicks off query

func GetMultiComments(prodId string) (*tablestore.GetRangeResponse, error) {

fmt.Println("batch get row started")

rangeRowQueryCriteria := &tablestore.RangeRowQueryCriteria{}
rangeRowQueryCriteria.TableName = TableName

startPK := new(tablestore.PrimaryKey)
startPK.AddPrimaryKeyColumn("ProdId", prodId)
startPK.AddPrimaryKeyColumnWithMinValue("CommentId")

endPK := new(tablestore.PrimaryKey)
endPK.AddPrimaryKeyColumn("ProdId", prodId)
endPK.AddPrimaryKeyColumnWithMaxValue("CommentId")

rangeRowQueryCriteria.StartPrimaryKey = startPK
rangeRowQueryCriteria.EndPrimaryKey = endPK
rangeRowQueryCriteria.MaxVersion = 1
rangeRowQueryCriteria.Limit = 1
rangeRowQueryCriteria.Direction = tablestore.FORWARD

getRangeRequest := &tablestore.GetRangeRequest{}
getRangeRequest.RangeRowQueryCriteria = rangeRowQueryCriteria

client := CreateClient()
getRangeResp, err := client.GetRange(getRangeRequest)	//panic

return getRangeResp, err

}

I refer to below SDK example:
https://github.com/aliyun/aliyun-tablestore-go-sdk/blob/master/sample/MultipleRowOperation.go
https://help.aliyun.com/document_detail/52265.htm

设置后，得到结果还是全部 column

Please support go module

运行的代码如下：

    searchRequest.SetTableName(tableName)
    searchRequest.SetIndexName(indexName)
    query := &search.GeoDistanceQuery{} // 设置查询类型为GeoDistanceQuery
    query.FieldName = "Col_GeoPoint"
    query.CenterPoint = "5,5"  // 设置中心点
    query.DistanceInMeter = 10000.0 // 设置到中心点的距离条件，不超过10000米
    searchQuery := search.NewSearchQuery()
    searchQuery.SetQuery(query)
    searchQuery.SetGetTotalCount(true)
    searchRequest.SetSearchQuery(searchQuery)
    // 设置返回所有列
    searchRequest.SetColumnsToGet(&tablestore.ColumnsToGet{
        ReturnAll:true,
    })
    searchResponse, err := client.Search(searchRequest)
    if err != nil {
        fmt.Printf("%#v", err)
        return
    }
    fmt.Println("IsAllSuccess: ", searchResponse.IsAllSuccess) // 查看返回结果是否完整
    fmt.Println("TotalCount: ", searchResponse.TotalCount) // 匹配的总行数
    fmt.Println("RowCount: ", len(searchResponse.Rows))

场景：1、在实例上创建表 2、go客户端连接实例操作数据。

在Sample代码中，遇到两个问题：

1、插入数据时如果不添加pk主键，不返回内容，也不报错，数据库也没有内容
2、客户端数据类型和实例数据类型不一致会导致客户端崩溃，报以下错误：
[HTTP Server] http: panic serving 127.0.0.1:62204: [tablestore] invalid input goroutine 43 [running]: net/http.(*conn).serve.func1(0xc000294aa0)
例如：主键或者某一列在实例中是int(11)，在客户端是int8

这里面的读，对与int/double等类型，没有读/跳过 len字段．　我查这个repo中，对于ColumnValue写，也奇怪的多写len＋type+value的值，和读根本不对应．．

不知道是以ｃｏｄｅ中这种非对称的写法是对的，还是按照plainbuffer的规范定义操作是对的．唉．

请问官方demo使用多元索引并发导出的时候先调用ComputeSplits方法获得并行度，这个方法返回的是1，这样还是一个并行度导出数据的。

针对官方文档提供的条件更新接口：https://help.aliyun.com/document_detail/35194.html

请问此SDK有封装这个接口吗？能否提供一个代码案例？

谢谢

Currently the maxium retry interval is 320ms, a constant, and in some cases, if it can be configured, it'll be better.

timeline 中的Scan函数，MaxCount参数没用，返回数据也无序。这种基本的问题(几行代码搞定)为何迟迟没人解决？？用户怎么使用？？

#45
#48

https://go.dev/wiki/Modules#releasing-modules-v2-or-higher

PutColumnWithTimestamp
提供一个example

func (rowchange *UpdateRowChange) PutColumnWithTimestamp(columnName string, value interface{}, timestamp int64) {
	column := &ColumnToUpdate{ColumnName: columnName, Value: value}
	column.Timestamp = timestamp
	column.HasTimestamp = true
	rowchange.Columns = append(rowchange.Columns, *column)
}

RangeRowQueryCriteria.Limit = 3 ， 实际仍然输出超过3行记录

000569f5-e56a-9201-4ac3-980a0ee23a01
{
"PrimaryKey": {
"PrimaryKeys": null
},
"Columns": null,
"ConsumedCapacityUnit": {
"Read": 0,
"Write": 0
}
}

使用TmLine.scan 获取timeline的数据，发现是无序的。
看了一下SDK的代码timeline/store.go返回的是一个map， timeline.go中的asyncScan直接循环获取map中k,v 导致结果无序

I tried to update MaxVersion without changing TimeToLive's value, but it hit 'OTSParameterInvalid TimeToLive cannot be 0 or less than -1.' issue with the code below.

	fmt.Println("UpdateTableSample started")
	updateTableReq := new(tablestore.UpdateTableRequest)
	updateTableReq.TableName = tableName
	updateTableReq.TableOption = new(tablestore.TableOption)
	updateTableReq.TableOption.MaxVersion = 5

	_, err := client.UpdateTable(updateTableReq)

当BatchGetRow中某个表的某个主键不存在时，会返回unexpect io end

在ots的文档中说到（https://help.aliyun.com/document_detail/27310.html?spm=5176.doc52265.6.615.NDnOUA）：
如果某行不存在或者某行在指定的 columns_to_get 下没有数据，仍然会在 TableInBatchGetRowResponse 中有一条对应的 RowInBatchGetRowResponse，但其 row 下面的 primary_key_columns 和 attribute_columns 将为空。

unexpect io end应该就是读到了空的primary_key_columns 和 attribute_columns。
应该判断一下错误类型，而不是直接返回unexpect io end。

/*if key.Option != nil {
keyOption := PrimaryKeyOption(key.Option)
responseTableMeta.SchemaEntry = append(responseTableMeta.SchemaEntry, &PrimaryKeySchema{Name: key.Name, Type: &keyType, Option: &keyOption })
} else {
responseTableMeta.SchemaEntry = append(responseTableMeta.SchemaEntry, &PrimaryKeySchema{Name: key.Name, Type: &keyType })
}/

TableMeta:

{
"TableMeta": {
"TableName": "auto",
"SchemaEntry": [
{
"Name": "uid",
"Type": 2,
"Option": null
},
{
"Name": "pid",
"Type": 1,
"Option": null
}
]
},
"TableOption": {
"TimeToAlive": -1,
"MaxVersion": 1
},
"ReservedThroughput": {
"Readcap": 0,
"Writecap": 0
},
"StreamDetails": {
"EnableStream": false,
"StreamId": null,
"ExpirationTime": 0,
"LastEnableTime": 0
}
}

写入cellname的时候，按规范应该要写入类型string．这里没有写入，难道服务端做了同样的忽略？不然应该读不到正确的cellname啊

`
panic: runtime error: invalid memory address or nil pointer dereference
[signal 0xc0000005 code=0x0 addr=0x0 pc=0x7509e5]

goroutine 1 [running]:
github.com/aliyun/aliyun-tablestore-go-sdk/tablestore.(*TableStoreClient).CreateTable(0xc000122000, 0xc0000cbeb0, 0x4, 0x4, 0x5)
D:/GoLang/GoWork/src/github.com/aliyun/aliyun-tablestore-go-sdk/tablestore/api.go:348 +0x965
main.CreateTable(0xc000122000, 0x31, 0x84e146)
D:/GoLang/GoWork/src/local_no_vcs/OTSTest202103040238/main.go:42 +0x5ef
main.main()
D:/GoLang/GoWork/src/local_no_vcs/OTSTest202103040238/main.go:17 +0xad
`

相关的上下文（github.com/aliyun/aliyun-tablestore-go-sdk/tablestore/api.go 348行附近）：
`
346 req.ReservedThroughput = new(otsprotocol.ReservedThroughput)
347 req.ReservedThroughput.CapacityUnit = new(otsprotocol.CapacityUnit)
348 req.ReservedThroughput.CapacityUnit.Read = proto.Int32(int32(request.ReservedThroughput.Readcap))
349 req.ReservedThroughput.CapacityUnit.Write = proto.Int32(int32(request.ReservedThroughput.Writecap))

`

目前TableStore Golang SDK的操作特别繁琐，代码量较多，请考虑支持ORM框架的实现，比如gorm。

Scenario:
Put a row into TableStore, expect to get the primary keys back since CommentId (2nd PK) is auto incremental int64

Bug: failed to get non-nullable "PrimaryKeys" from PUTRow

Exception: put response: {"ConsumedCapacityUnit":{"Read":0,"Write":1},"PrimaryKey":{"PrimaryKeys":null},"RequestId":"0005a22b-94b8-67ca-8ac0-dc0b38e2f4bc"}

Justification:
according to the api at line 649, https://github.com/aliyun/aliyun-tablestore-go-sdk/blob/master/tablestore/api.go; PrimaryKeys should be returned.
according to doc, https://www.alibabacloud.com/help/doc-detail/27306.htm, The data type after the row is successfully written. Currently, only the primary key can be returned, which is used for the auto-increment function of the primary key column.

Coding Language: GOLANG

type Comment struct {
ProdId string // prodid as partition key
CommentId int64 //row key for each comment
Content string //comments text
}

CreatTable PK meta
meta := new(tablestore.TableMeta)
meta.TableName = "commentTest"
meta.AddPrimaryKeyColumn("ProdId", tablestore.PrimaryKeyType_STRING) // prodid as partition key
meta.AddPrimaryKeyColumnOption("CommentId", tablestore.PrimaryKeyType_INTEGER, tablestore.AUTO_INCREMENT) //row key for each comment
meta.AddDefinedColumn("Content", tablestore.DefinedColumn_STRING) //comments text

method

putRowChange := new(tablestore.PutRowChange)
putRowChange.TableName = "commentTest"

meta := new(tablestore.PrimaryKey)
meta.AddPrimaryKeyColumn("ProdId", c.ProdId)
if c.CommentId > 0 {
	meta.AddPrimaryKeyColumn("CommentId", c.CommentId)
} else {
	meta.AddPrimaryKeyColumnWithAutoIncrement("CommentId")
}
putRowChange.PrimaryKey = meta
putRowChange.SetCondition(tablestore.RowExistenceExpectation_IGNORE)

putRowChange.AddColumn("Content", c.Content)

request := new(tablestore.PutRowRequest)
request.PutRowChange =putRowChange

client := CreateClient()
resp, err := client.PutRow(request)

data, _ := json.Marshal(resp)
fmt.Println("put response: ", string(data))

It's a common requirement that user may ensure the table exists, if not, EnsureTable should create table automatically, something like Sync in timeline pkg

目前go版这个timeline模型可以正常跑不?
sync表和store表需要自己创建吗?
我按readme要求, 设置了配置文件. 但是无法跑起来.
返回"OTSParameterInvalid Invalid table name:...."错误.

如果需要自己创建表. 相关的主键字段如何设置呢?

t := reflect.TypeOf(value)
switch t.Kind() {
case reflect.String:
v.Type = otsprotocol.PrimaryKeyType_STRING
case reflect.Int64:
v.Type = otsprotocol.PrimaryKeyType_INTEGER
case reflect.Slice:
v.Type = otsprotocol.PrimaryKeyType_BINARY
default:
panic(errInvalidInput)
}
只支持string/int64/slice，导致如put response或者stream record中携带的primaryKey，用户都没有办法直接用，因为primaryKey中的各个column的value都是interface类型的，用户需要拷贝处理一次才能继续使用，eg:
row := getAStreamRecord() //type *ots.StreamRecord
putChange.PrimaryKey = row.PrimaryKey // panic!
需要改写成
pk := new(ots.PrimaryKey)
pk.AddPrimaryKeyColumn("pk_0", int64(row.PrimaryKey.PrimaryKeys[0].Value.(uint64)))
putChange.PrimaryKey = pk

func BatchWrite(cli *tablestore.TableStoreClient, tableName string) {
	cs := make(map[string][]tablestore.RowChange)
	for i := 0; i < 200; i++ {
		putRowChange := &tablestore.PutRowChange{TableName: tableName, PrimaryKey: &tablestore.PrimaryKey{}}
		putRowChange.PrimaryKey.AddPrimaryKeyColumn("pk1", strconv.Itoa(i))
		putRowChange.PrimaryKey.AddPrimaryKeyColumnWithAutoIncrement("pk2")
		putRowChange.AddColumn("col_"+strconv.Itoa(1), strconv.Itoa(i))
		putRowChange.AddColumn("col_"+strconv.Itoa(2), int64(i))
		putRowChange.AddColumn("col_"+strconv.Itoa(3), strconv.Itoa(i))
		putRowChange.AddColumn("col_"+strconv.Itoa(4), int64(i))
		putRowChange.AddColumn("col_"+strconv.Itoa(5), strconv.Itoa(i))

		putRowChange.SetCondition(tablestore.RowExistenceExpectation_IGNORE)
		cs[putRowChange.GetTableName()] = append(cs[putRowChange.GetTableName()], putRowChange)
	}

	res, err := cli.BatchWriteRow(&tablestore.BatchWriteRowRequest{RowChangesGroupByTable: cs})
	if err != nil {
		fmt.Println("batch request failed with:", err.Error())
	}
	fmt.Println(res)
}

debug screenshot