On 1/1/07, Alexei Zakharov <[EMAIL PROTECTED]> wrote:
Hi Andrew,
IMHO if exception is thrown then the object is in inconsistent state
anyway. So it does not really matter what the length() method returns.
However, negative length also doesn't make sense to me. So here is my
+1 for not touching the length.
As my understanding, such exception is a way to tell user that the parameter
is invalid. But the object can still be used later. Therefore, I'd like not
to change the length value here.
Now we've got incompatibility almost in every SerialBlob method, right?
yes. And maybe one more in setBinaryStream method. :)
Regards,
2007/1/1, Andrew Zhang <[EMAIL PROTECTED]>:
> Hi folks,
>
> I encountered another compatibility issue (the fourth one?) of
SerialBlob
> class. It's about truncate(long length) method. Spec says "Throws
> SerialException - if there is an error accessing the Blob value; or the
> length to truncate is greater that the SerialBlob length". But it
doesn't
> mention what happens if the length is negative. I tried following test
code:
>
> public void testTruncateJ() throws Exception {
> byte[] buf = { 1, 2, 3, 4, 5, 6, 7, 8 };
> SerialBlob serialBlob1 = new SerialBlob(buf);
> try {
> serialBlob1.truncate(-1);
> } catch (SerialException e) {
> System.out.println(e.getMessage());
> }
> System.out.println(serialBlob1.length());
> }
> The RI output is:
> Invalid arguments: position cannot be less that 1
> -1
> The first output SerialException is quite reasonable and the message
also
> sounds good. But the length is then changed to -1. IMO, if truncate
method
> regards -1 as an illegal argument, it should not change the length value
> anyway. So I suggest that Harmony throws SerialException as RI and
doesn't
> change the length value (any internal data of serial blob). What's your
> opnion? Thanks in adavance!
>
> --
> Best regards,
> Andrew Zhang
>
>
--
Alexei Zakharov,
Intel ESSD
--
Best regards,
Andrew Zhang
