Closed GoogleCodeExporter closed 9 years ago
I assume you mean directly storing json documents as byte arrays, without going
through an intermediate object conversion.
I think this is very useful indeed, in particular when the object<->json
conversion is externally done.
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 1:13
Sven,
what's your take on this?
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 1:14
Yes, indeed. yes, like having an API like:
put(byte[] key, byte[] value)
Original comment by grec...@gmail.com
on 16 Dec 2010 at 1:16
It's a rather small thing to do.
I don't see any obvious use case for referring to the key using a byte-array,
though.
Unless there are any concrete reasons for this I suggest:
put(String key, byte[] value).
Assigning it to myself.
Original comment by johansso...@gmail.com
on 16 Dec 2010 at 1:49
Good point about the key: no need to use byte arrays for that.
Anyways, I think we should provide alternative bytearray-based methods for all
read/write operations, rather than just put ... thoughts?
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 1:51
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 1:52
The use-case I'm referring to is providing a persistent storage backend for
akka based on TS.
In this case the key is passed as a byte array.
Original comment by grec...@gmail.com
on 16 Dec 2010 at 2:01
David,
yo should manually convert the key from byte[] to utf8-string prior to passing
it to the client: it shouldn't be a problem, nor provide any performance
penalty.
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 2:04
Yes, no problem, I wanted only to give you an example of a use-case. However,
I agree on the fact that the key as a String is a more general use-case
Original comment by grec...@gmail.com
on 16 Dec 2010 at 2:08
Sven,
just thinking aloud about possible implementations.
We have two ways to do that:
1) Add alternative byte[]-based methods, mirroring already existent
object-based ones.
2) Leave already existent interfaces untouched, and implement a "RawJsonObject"
which users willing to pass raw json data back and forth will have to use to
wrap their json byte[]: our client will recognize the use of such a special
object and apply no conversion but simply using the wrapped json byte[].
Thoughts?
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 2:09
As for using a byte-array as key, that Akka somehow works this way does not
feel like a general enough use case to add a method to the API for this.
Come to think of it, though, given that you pass in a byte-array value that in
fact _is_ a valid UTF-8 JSON String, it should be possible to do this already,
just as you can pass a JSON String.
I'd need to double-check this, but I'm pretty sure.
Original comment by johansso...@gmail.com
on 16 Dec 2010 at 2:59
Sven,
I agree. I really don't mind to convert the byte array into a String. Really no
problem
Original comment by grec...@gmail.com
on 16 Dec 2010 at 3:03
Sorry, David - I didn't see comment(9) above about that.
Original comment by johansso...@gmail.com
on 16 Dec 2010 at 3:09
Please go ahead with the discussion on the project mailing list:
http://groups.google.com/group/terrastore-discussions/t/a04a72880ab3c005
Original comment by sergio.b...@gmail.com
on 16 Dec 2010 at 3:29
Closing this issue, as no change is required.
Original comment by johansso...@gmail.com
on 17 Dec 2010 at 9:59
Original issue reported on code.google.com by
grec...@gmail.com
on 16 Dec 2010 at 1:07