Need official interpretation - java.io.Reader read(char[] cbuf) method

Question

Hello Everyone,
I need Sun API's official interpretation on java.io.Reader read(char[] cbuf) method;
Say, I provide this read method a char[] of size 8096, could the implementing class return even before reading into the whole buffer array? Say, the implementing class just reads 1024 chars and return 1024 as the method output, has the implementing class fulfilled its obligation?
In my mind, the read method should attempt to read and fill the buffer array until of course i/o error occurs or end of stream is reached? See BufferedReader source for this type of implementation.
Or could the read method just attempt to read into the buffer (with no obligation) and can return before completely filling the buffer?
Sincere Regards.

Answer 1

This is the situation: (and we are caught in between arbitrating on whom to ask to make the fix)
Please note that I am not taking any names here, but remember these vendors are the biggest in their market.
Vendor Db A has an API wherein it returns a stream and its length;
Vendor B consumers this API via the java.io.Reader class as follows:
char[] data = new char[(int) VendorAStream.length()];
VendorAStream.read(data);
Vendor B contends that per java.io.Reader API, it is responsibility of Vendor A to implement the read method correctly!!! (which is to read the data completely, not partially); Vendor B further contends that their code works fine with all the other Db vendors (except Vendor A); [Quite *tupid reasoning in my mind ;-) ]
Vendor A says per java API, we can partially fill the buffer (they are specifically only reading 1024 chars from the stream)
Initially, I thought, Vendor B must fix the code to handle partial reads. But no where it is mentioned clearly that Reader.read only attempts to read the chars and can return before filling the buffer.
Dannyyates,
(*My interpretation*) The reason it says "*maximum* number of characters to read" is for the situation when end of stream is reached before fully filling the buffer. In that case, the length of buffer and number of actually chars read wont match. This explanation covers for your observation (c)
BIJ001, No problem in your explanation. I understand and I am merely pointing you to alternate interpretation of the API.
Please comment. Thanks for your time to discuss this.
Sincere Regards.