Serializing a DataSet with empty data tables loses all column information

We have introduced some optimisations at work to a web service to improve the speed. The aim here is to serialize some data to a file to avoid going to the database if we have already made a call to the same Stored Proc with the same arguments within a cache timeout window

ds = DataAccess.ExecuteDataSet

Try
   If fileCache.CacheFilePath IsNot Nothing AndAlso Not File.Exists(fileCache.CacheFilePath) Then
          ds.Serialize(fileCache.CacheFilePathTemp)
          fileCache.RenameTemp()
    End If
Catch ex As Exception
    ' couldn't write to the cache? Don't want this to affect the report, so just write to the output
    ' and move on
    Debug.Print("Could not write to report cache: " + ex.Message)
End Try

Now consider that the DataSet contains two tables, but the tables are empty When serialized, we get the following JSON

{
  "Table": [],
  "Table1": []
}

We have lost the following information

• The name of the table
• The name of any columns and their associated data type

The fix is as follows

1. Before outputting the Rows for each table, output the table name and columns/column data types
2. When reading the data back, add the columns and set the table name before attempting to create the rows.

[harrisonbs]

With all due respect, I don’t think this issue should be closed. It should be fixed in the manner I said above - use a setting to determine if you want to preserve column information. Serializing the ROWS data only can lead to a lossy deserialization which caused a serious issue for my client because the object that was serialized and then deserialzed again was NOT equal to the original object.

Serialization is the process of converting an object into a stream of bytes in order to store the object or transmit it to memory, a database, or a file. Its main purpose is to save the state of an object in order to be able to recreate it when needed.

https://docs.microsoft.com/en-us/dotnet/csharp/programming-guide/concepts/serialization/

Saying it’s supposed to be near enough is good enough is not actually good enough. And if I have a solution, why wouldn’t you include it back into the code in the manner highlighted?

[robinsmh]

I can’t believe that this is still a problem - but it is. Even more so in ASP .Net Core where use of the session state requires everything to be serialised.

We don’t serialise to json because we want ‘nice’ clean json, we serialise because we want a json representation of ALL the relevant data. For a datatable, this must absolutely include the column names and types even if there are no rows, and it must be reversible back to an equivalent datatable when de-serialized.

PLEASE PLEASE PLEASE add an option to include the table schema in the json, otherwise what’s the point of supporting DataSet/DataTable at all? It it simply not fit for purpose currently.