Suggestion: Replace `TableNamePrefix` with `ITableNameFormatter` for Better Flexibility

[icanhasjonas]

 Currently, the API uses a TableNamePrefix string to modify table names. This works, but is inflexible.

Proposal

Introduce an ITableNameFormatter interface to handle table name resolution.
This allows for:

• Custom naming strategies (e.g., environment-based, hashing, localization)
• Cleaner separation of concerns
• Easier testing and mocking

Migration Path:

• Mark TableNamePrefix as [Obsolete]
• If TableNamePrefix is set, internally create a PrefixTableNameFormatter to maintain compatibility

This preserves backward compatibility while enabling greater extensibility.

[firenero]

Hi @icanhasjonas, thanks for the suggestion and submitting a PR. I'm happy to provide more flexibility for table naming. This approach introduces an API change that will eventually require all users to spend time migrating to the new API so I'd like to get a better understanding on the value this change brings.

My current concerns are:

• There is only one example of PrefixTableNameFormatter in the PR but I wonder what other use cases there are? It can be used for something simple like suffix-formatter but I'm not yet sure that it's worth introducing a breaking change. More complex scenarios like hashing or reading environment variables can introduce performance degradations if they are not cached because TryFormat is called for every request.
• There is already a way to provide a custom table name for requests by using WithTableName() builder methods. It obviously more hassle than setting it once but once again, I'm trying to avoid pushing any work on library users for a feature that they might not need.
• If we do the change, we should provide more predefined formatters in the library to justify the migration. Otherwise, there will be no new functionality out of the box.

[hugobritobh]

I encountered an issue that the ITableNameFormatter might be able to solve. I need to pass the table name dynamically. I used TableNamePrefix and had to modify the source code to set the default TableName to String.Empty.

This allowed me to pass the entire table name through TableNamePrefix, while keeping the default TableName empty. Of course, this was just a test to work around the problem.

[firenero]

@hugobritobh Thanks for the example! One question - would it be simpler for you to use .WithTableName("<your_full_table_name>") builder methods to override table name instead of modifying sources? I understand that it's less convenient than potential ITableNameFormatter because you have to apply it to every query. But I would like to know if this approach functionally covers everything that ITableNameFormatter provides except the convenience of usage.

[hugobritobh]

In my example, the approach you presented would work and would be simpler.