fogfish / dynamo Goto Github PK

https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/LSI.html
https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/GSI.html

The projection expression allows to limit number of columns. These columns are deductible from the type definition.

The following syntax does not protect api from mixing constrains of different types

type Constrain[T Thing] interface{}

It is required

type Constrain[T Thing] interface{ TypeOf(T) }

Identity function is implemented as

func (x *myType) Identity() (string, string) {
	return curie.IRI(x.ID).String(), curie.IRI(x.Suffix).String() 
}

String encoding preserves : (e.g. u:abc). It becomes difficult to handle as S3. S3 requires conversion of type to path (e.g. u/abc/).

Earlier release used LastEvaluatedKey to build a sequence of elements. New version migrated to Slice. It complicates the pagination. Implement the concept of cursor to fix an issue.

Usage of dynamo.ID is an approach to label a struct. However, it brings few practical issues:

1. It creates deps between core types and aws sdk due to dynamodb.AttributeValue type
2. Usage of type composition does not make an explicit mapping of core ID to dynamo ID

Consider a following scenario:

type A struct {
  ID curie.IRI `dynamodbav:"-"`
  Name string `dynamodbav:"name,omitempty"`
}

type dbA struct { A }

func (x dbA) MarshalDynamoDBAttributeValue(av *dynamodb.AttributeValue) error {
   return someHelperFunction(x.A.ID, x)
}

func (x dbA) UnmarshalDynamoDBAttributeValue(av *dynamodb.AttributeValue) error {
   return someHelperFunction(&x.A.ID, &x)
}

This approach still do not address an issue of linked data

type A struct {
  ID curie.IRI
  Link *curie.IRI
}

1. Put the document to s3 using Put

{
   "a": "test",
   "list": [{"id":1, "text": "a"}, {"id":2, "text": "b"}, {"id":3, "text": "c"}, {"id":4, "text": "d"}]
}

2. Update the document at s3

{
   "a": "test",
   "list": [{"id":5, "text": "e"}, {"id":6, "text": "f"}]
}

after the update, elements of struct at list is corrupted.

// FMap transforms sequence
func (seq *dbSeq) FMap(f func(Gen) error) error {
	for seq.Tail() {
		if err := f(seq.slice.Head()); err != nil {
			return err
		}
	}
	return nil
}

return nil if seq.Tail() seeding fails with an error

ValidationException: One or more parameter values are not valid. The AttributeValue for a key attribute cannot contain an empty string value. Key: prefix

dynamo.Decode does not handle error gracefully
https://github.com/fogfish/dynamo/blob/main/id.go#L146

https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/Expressions.UpdateExpressions.html

Allow user to pass the hash key sort key as config values.

See official doc for details https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/HowItWorks.ReadConsistency.html

Existing library interface tries to mimic a general purpose dynamo io. However, it is only designed to support adj. list pattern. It benefits if API emphasis this facts.

• Thing interface to explicitly support Partition / SortKey
• Stream interface is a file-system interface
• Helper methods to deal with I/O errors, especially NotFound one.