Hi yang, how about the issue? since the cube is built day by day, it the cube merge feature doesn't work, it will affect the performance if the count of the segments is big according to the documents~
2015-03-16 13:14 GMT+08:00 Li Yang <[email protected]>: > Interesting.. I'll give a look in the afternoon > > On Tue, Mar 10, 2015 at 8:56 PM, dong wang <[email protected]> wrote: > > > HI yang, as the log above, it throws this exception from there: > > System.arraycopy(bytes, 0, returnValue, offset, bytes.length); and > > as debugged, the value of "bytes" is "10000-01-01", so length of "bytes" > is > > 11, and the value of "returnValue" is "9999-12-31", whose length is 10, > > therefore, the exception is thrown out. what's more, when this exception, > > the parameter "id" passed to the following function is 3652427, then > > function getValueBytesFromIdImpl() will internally call > > getDateFromNumOfDaysSince0000() as below, and, the returned date is > > "10000-01-01", so, is it possible that there is something that we haven't > > taken it into consideration for function > "getDateFromNumOfDaysSince0000()"? > > > > final protected int getValueBytesFromIdImpl(int id, byte[] returnValue, > int > > offset) { > > String date = getValueFromId(id); > > byte bytes[]; > > try { > > bytes = date.getBytes("ISO-8859-1"); > > } catch (UnsupportedEncodingException e) { > > throw new RuntimeException(e); // never happen > > } > > System.arraycopy(bytes, 0, returnValue, offset, bytes.length); > > return bytes.length; > > } > > > > > > private Date getDateFromNumOfDaysSince0000(int n) { > > long millis = ((long) n - 719530) * 86400000; > > return new Date(millis); > > } > > >
