[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Newsgroup Home]
|
[news.eclipse.technology.eilf] Re: [Xml Storage] API :: first draft
|
Hi,
some comments inline ...
Am 06.05.2008, 13:31 Uhr, schrieb tmenzel <tmenzel@xxxxxxx>:
another item i might want to put here is the client's default collection
if we have more than one and these are not transparent to the user.
collections, as far as i can tell, are not 100% defined as to what they
represent in each implementation of Xquery but may be used to represent
diff. data stores or SEGEMNTs of the same store, etc. and that is smth.
we might want to control on the client side, e.g. as a private temp
storage etc.
on the other hand we might decide to put that one level higher and make
that part of the config, such that one service only represents one
collection and u would have to connect to another service that
represents another collection.
I think it would be OK to have the several collections of one XML database
accessed via a single XSS service, it would make implementation of clients
and service deployment easier: One XSS service instance represents one
physical XML storage (server? I'm not quite sure about the terminology
yet) and everything in this storage can be accessed via this service.
If so: Would it make sense (be possible?) to have some collection
"management"
methods in XmlStorageService? E.g. getAvailableCollections,
createCollection,...?
as u can see: i'm not quite sure on the subject
Hope this helps to make up your mind (-:
- About the XssConnection methods: I wonder if we could need bulk
mehods like addRecords(Record[]), or simply make the existing methods
take a variable number of arguments, e.g. addRecord(Record...) to be
able optimize performance?
possible, but i tried to resist the temptation of offering too many
convenience methods at this time. once the API is stable and we got some
experiences, we may add such methods. before that, adding some helper
classes should be the way to go IMHO.
That's OK for the moment. Just wanted to add that these would not be pure
convenience methods that could be moved to helper classes but that they
should really optimize something when talking to the actual database
server.
Cheers,
Juergen.