Data binding facade design bug = delete ambiguous for PojoRepository<T, String>

[marklogic]

Data binding facade design bug = delete ambiguous for PojoRepository

from java-client-api.

[sammefford]

I was going to change it to deleteAll(String... collections), but I see now I can't do that since the REST API doesn't support multiple collections on delete. If I provided deleteAll(String collection) I could delete any document in the database in that collection, but I expect the pojo facade to limit to only the types of documents managed by the pojo facade, and not delete documents unmanaged by the pojo facade whether they're in the matching collection or not. The best way is to filter to documents in a collection added by the pojoFacade but I can't provide two collection args to the REST DELETE /search endpoint. So for now I'm limiting this down to deleteAll() and it will just send the one collection auto-added by the pojo facade. This has the advantage of never deleting a document not managed by the pojo facade, and facilitating deletion of all docs managed by the pojo facade.

I'll mark this issue as closed. Please re-open if you have a concern about this solution.

[collwhit]

Sam, if you need changes at the REST layer to support what you need to do in the Java interface, we should be talking about that. I think your conclusion is reasonable; just saying that you shouldn't assume you can't get an enhancement if you need it.

[sammefford]

I would like the enhancement. I added an RFE.

[grechaw]

If we don't implement this, then people cannot use the pojo facade with objects that use String IDs. I think that's pretty serious actually! I like Sam's solution to the problem. Do you think it's just too late to fix for 8.0-1? I need to change my implementation in Samplestack (just a little though) if we're not fixing this.

[sammefford]

Charles, I'm not understanding your comment. I added a deleteAll() method with no args that deletes all docs for the class managed by the PojoRepository. We still have the delete(ID id) method. So the only thing we're waiting for is deleteAll(String... collection) which I consider a convenience, not something that prevents people from using the pojo facade.

[grechaw]

Sorry I misunderstood the work already done. In that case I'm in good shape! The important thing is that delete(String...) doesn't exist anymore

[sammefford]

Yeah, I should have clarified that delete(String...) doesn't exist anymore. I agree that was ambiguous. I'm confused why the Java compiler allows me to get away with such ambiguity. But thanks for finding it so I could fix it!

[grechaw]

Well to be fair the "ambiguous" problem WAS a compiler error ... but only if your class uses String for .