For insertions and queries, use a unified interface

[jieguangzhou]

For MongoDB and Ibis
Currently:

• MongoDB uses insert_many for data insertion and find for queries.
• Ibis uses insert for data insertion and select for queries.

For these very frequent interfaces, we can provide a unified method:

db["table_or_collection_name"].insert(datas)
df["table_or_collection_name"].query(a=1, b=2)

• insert_one
• query_one
• filter
• limit
• output_fields

By standardizing these common operations, similar to the most frequently used ones in superduperdb, everything will become simpler to use. For complex operations, the original logic can still be used.

As for the names, we can discuss and define query/select/find later.

[blythed]

That's nice, but I feel that it might make messaging difficult. Until now our USP has been "use your frameworks"
exactly as you are used to. Now we add a new uniform abstraction. Although, personally this is a feature I would enjoy.

Later when we index things which aren't directly websites, then we can do things like:

db = superduper("http://bbc.com")

db["europe"].query(tag='a')      # points to http://bbc.com/europe, selects link tags

[jieguangzhou]

This is not conflicting, they can also use the original method, WDYT?

If we don’t want to expose it to users, we can choose not to expose this method.

However, this method is of great help to our UT and integration testing. We can now test two types of databases with one set of code.

For example, These two test cases have a lot of redundant codes

[blythed]

Yes, I see your point, could be potentially very useful, yes.

[fnikolai]

@jieguangzhou

Since query is an abused term that may also include actions such as Create() or Delete(), I 'd prefer select().

[jieguangzhou]

@jieguangzhou

Since query is an abused term that may also include actions such as Create() or Delete(), I 'd prefer select().

Sure, used select now