Browsing the database, detecting bindings

Question

Am I correct in my understanding that there is no way to detect the binding that should be used to deserialize the entry?
I would like to write a generic database browser but it doesn't seem possible. I suppose I could support only databases that catalog their bindings in a standard way but that seems rather limiting.
A major hurdle for some people feeling comfortable with berkeleydb is the ability to browse the data. I would like to write a developer tool to help in this regard. Maybe one already exists but I did not find any. Your input would be greatly appreciated.

Answer 1

Hi,
Am I correct in my understanding that there is no way
to detect the binding that should be used to
deserialize the entry?Correct. In Berkeley DB in general, the bindings and other metadata are not stored in a database, they are typically represented only in code or in an application-specific form. For many users, the ability to completely control their own schema is considered a strength, but for other users having a built-in schema is desirable.
I would like to write a generic database browser but
it doesn't seem possible. I suppose I could support
only databases that catalog their bindings in a
standard way but that seems rather limiting.Yes, a generic browser that would work with any BDB database would have to either 1) only display raw data bytes, which is not much better than DbDump, or 2) require some sort of standard schema information, as you say. One of the problems in implementing (2), is that you must decide what type of schema you wish to support -- objects and classes in a Java sense, flat records with simple fields, arbitrary XML-based structure, etc. I don't think any one type of schema will work for all BDB users.
A major hurdle for some people feeling comfortable
with berkeleydb is the ability to browse the data. I
would like to write a developer tool to help in this
regard. Maybe one already exists but I did not find
any. Your input would be greatly appreciated.I claim that most users who would like a built-in browser would also like a built-in schema facility, which is exactly what we provide in the Direct Persistence Layer (persist package). The DPL was built with the idea of adding a browser in mind, but we haven't done it. If you want to implement a browser, I encourage you to implement a DPL-based browser. The DPL exposes all metadata and a browser could be built using the RawStore class.
What form of UI for a browser is most desirable to you?:
- browser-based, via a servlet perhaps
- IDE-based, as an Eclipse, NetBeans, or JBuilder plugin perhaps
- standalone Swing app
Are you interested in browsing the data, the metadata, or both?
--mark