RE: [Tm4j-developers] Index interface proposal checked in

[Kal A. <ka...@te...>]

At 21:02 28/02/2002 +0100, Florian G. Haas wrote:
>Kal,
>
>here are a few comments.
>
>1. The fact that the methods returning an IndexMeta object are called
>getIndexMeta() in IndexProvider and getIndexMetaData() in IndexManager seems
>like an inconsistency to me.

Agree - will rename to getIndexMeta() as this matches the class name.

>2. Perhaps IndexManager.hasIndex(String) should be renamed to
>isRegistered(String) or isRegisteredIndex(String). In addition, I think that
>since already have IndexMeta.isAutomaticallyUpdated(), we should also have
>an isAutomaticallyOpened() replacing requiresOpen() (with return values
>reversed, of course).

Good suggestions!

>3. I think that IndexProvider.getIndexNames() should be renamed to
>getIndexClassNames(), or perhaps be amended/replaced by a getIndexClasses()
>method returning a Class[] array. A boolean isSupportedIndex(Class
>indexClass) method may also be helpful. Or one might go the whole nine yards
>and overload close(), getIndex(), getIndexMeta() and open() in such a way
>that they also accept an argument of type Class, representing the index
>class. The same could be done to the methods in IndexManager.

I wanted to keep the IndexProvider interface as small and easy to implement 
as possible - it really only needs to expose methods for the IndexManager 
implementations to use. So I'll keep these suggestions in my back pocket 
for now...but I do think that overloading the IndexManager functions to 
accept Class parameters as well as String parameters makes sense.

>4. Index.reindex() documents that it throws UnsupportedOperationException
>which does not exist.

it is java.lang.UnsupportedOperationException.


>5. IndexProvider.getIndex() and IndexProvider.close() both document two
>exceptions which they do not throw.

Oops - my mistake - too eager to get the updates out :) They should both 
have a throws clause.

>6. IndexManager.getIndex() documents one exception which it does not throw.

Another "itchy commit finger" error ;-)


>Hope I'm making sense,


Definitely!  Thanks for taking the time to review and comment.

Cheers,

Kal