-
Notifications
You must be signed in to change notification settings - Fork 350
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fixes #3068: Throws ResourceTypeAnnotationNotFirst error when payload… #3069
base: main
Are you sure you want to change the base?
Conversation
… has both @removed and @odata.type
|
||
currentNode = currentNode.Next; | ||
} | ||
while (currentNode != this.bufferedNodesHead); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Was wondering how currentNode
ever be bufferedNodeHead
then I confirmed that this is a circular linked list. I think we should abstract traversal and access of the nodes through some helper method or inner class to avoid accidentally accessing the nodes the wrong way.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I checked the logics in 'BefferingJsonReader' and that's the way to check the end of link.
break; | ||
} | ||
|
||
currentNode = currentNode.Next; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
can currentNode.Next
ever return null
? Should account for that?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If we have the 'bufferedNodesHead', the 'Next' is not null because if it's empty, 'bufferedNodesHead.Next == bufferedNodesHead'.
|
||
if (currentNode.NodeType == JsonNodeType.Property && | ||
currentNode.Value.ToString() == propertyName && | ||
currentNode.Next != currentNode) // must have another node |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do check whether there's another node after this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just to see if I understand, property and value are stored in separate node, so if we find the expected property node, we want to make sure there's a value node after it. Is that correct?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
That's correct. Property and value are two node types and they are stored in separate node.
if (this.JsonReader.TryGetValueFromBuffered(ODataJsonConstants.PrefixedODataTypePropertyName, true, out value) || | ||
this.JsonReader.TryGetValueFromBuffered(ODataJsonConstants.SimplifiedODataTypePropertyName, true, out value)) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nit: make it clear what the true
arguments imply
if (this.JsonReader.TryGetValueFromBuffered(ODataJsonConstants.PrefixedODataTypePropertyName, true, out value) || | |
this.JsonReader.TryGetValueFromBuffered(ODataJsonConstants.SimplifiedODataTypePropertyName, true, out value)) | |
if (this.JsonReader.TryGetValueFromBuffered(ODataJsonConstants.PrefixedODataTypePropertyName, removeIfExists: true, out value) || | |
this.JsonReader.TryGetValueFromBuffered(ODataJsonConstants.SimplifiedODataTypePropertyName, removeIfExists: true, out value)) | |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
#Resolved.
@@ -114,6 +122,8 @@ internal void ReadResourceTypeName(IODataJsonReaderResourceState resourceState) | |||
|
|||
// Read the annotation value. | |||
resourceState.Resource.TypeName = this.ReadODataTypeAnnotationValue(); | |||
|
|||
// resourceState.TypeAnnotationFound = true; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why was this statement added then commented out? Should we remove it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yep, miss to clean it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
#Resolved.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The current design defeats the whole purpose of streaming -- for example, if the resource does not have an @odata.type we will read to the end.
@@ -81,6 +81,64 @@ internal BufferingJsonReader(IJsonReader innerReader, string inStreamErrorProper | |||
this.currentBufferedNode = null; | |||
} | |||
|
|||
internal bool TryGetValueFromBuffered(string propertyName, bool removeIfExist, out object value) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This could have some pretty severe performance ramifications.
If a payload doesn't have the value that we try to get, we will read the entire payload into memory. For very large, or streamed, responses, this could overflow memory.
This kinda defeats the whole point of ordering, which is that we can process the response in a streaming fashion.
The previous implementation assumes that type, if present, is the first thing we read. So we read the first node and, if it's type, we store that, and if it is not type, we assume there is no type and continue reading the payload as normal.
I would have expected we simply expand that logic. That is, we read the first node. If it is one of our "metadata" nodes (context, removed, type, id), then we get the value and repeat for the next node. As soon as we find a node that is not one of our metadata nodes, we break and start reading the payload as normal.
… has both @removed and @odata.type
Issues
This pull request fixes #3068.
Description
Make sure the reordering json reader reorder @context, @removed, @type, @id in this ordering.
@mikepizzo, Why does @removed should be property before @type?
We cannot assume the end customer is reading the payload using delta reader or normal entity reader, so, we don't know whether the @type position is the first property or second property if it contains @context. So, in this PR, try to retrieve the @type directly from the buffer. But, we should make sure:
a) Can a normal entity payload contain 'odata.removed'?
b) Can we allow use 'resource reader' to read the payload containing 'odata.removed'?
@mikepizzo
We can skip the value/validation if the 'odata.removed' has the wrong value. It's same concern that if we do like this solution?
Checklist (Uncheck if it is not completed)
Additional work necessary
If documentation update is needed, please add "Docs Needed" label to the issue and provide details about the required document change in the issue.