Welcome to Software Development on Codidact!
Will you help us build our independent community of developers helping developers? We're small and trying to grow. We welcome questions about all aspects of software development, from design to code to QA and more. Got questions? Got answers? Got code you'd like someone to review? Please join us.
Mixing "operational" database models with archiving ones in the database context
Context
Our product owner has realized that some entities are duplicated from time to time and that a merge is required. This should clean up existing duplicates and also allow special users to merge entities in the future.
The merge process is defined as follows (this should be done atomically):
- identify "to be merged" and "main" entities
- archive "to be merged" information
- parent most of the children of "to be merged" to "main"
- write some archiving metadata (who, when etc.)
- remove "to be merged" (with cascade)
The entity is rather used as it has a couple of dozens of other entities referencing it (foreign keys).
Dilemma
I am wondering how to proceed about this as I see two main options.
Entity Framework approach
- add all archive related models in the database context
- easily archive data by mapping existing models to archive ones
- all changes are done in a context transaction
Pros: rather easy to implement
Cons: bloats the context with archive related models (add a dozen models that have very similar names to the operational ones and might create confusion when having to work with DbSet related to these).
SQL approach
Create a stored procedure to tackle the archiving.
Pros: avoid bloating the context and app with archiving functionality
Cons: use SQL stored procedure which is harder to write and maintain.
Another way is to create another database context for archiving stuff, share the connection string and explicitly handle the transaction.
I am wondering if there is any best practice for such functionality.
1 answer
Generally speaking, switching data access technologies to avoid a naming conflict seems overkill. Doing that will increase the set of technologies contributors must be familiar with, and impede code reuse.
So, how to solve the similar names part? I'd probably choose names that are obviously different, for instance by requiring a big eye catchy prefix like ArchivedOrder. Using a prefix ensures that code completion will only suggest archive tables if a programmer explicitly types "Archive", and thus makes it very hard to accidentally refer to the wrong table.
Other, entity framework specific, solutions may exist, but alas I am not familiar enough with EF to advise you on their applicability.
1 comment thread