Phil, I did not see you provided a webrev: looks good.
I will run the test against marlin and check if it works and also if it test exceeding its initial array capacity. Thanks and sorry for the noise, Laurent Le 2 avr. 2015 07:17, "Laurent Bourgès" <[email protected]> a écrit : > > Hi Phil, > > I can fix that bug in pisces as I already did it in Marlin. > > You can have a look to Marlin Dasher as your proposed change look very similar to what I did. > > Laurent > > Le 2 avr. 2015 00:58, "Phil Race" <[email protected]> a écrit : > > > > https://bugs.openjdk.java.net/browse/JDK-8075942 > > http://cr.openjdk.java.net/~prr/8075942/ > > > > It appears that the amount to grow the array is not taking into account > > that before the System.arraycopy we first store 'type' and increment the > > offset :- > > > > 150 firstSegmentsBuffer[firstSegidx++] = type; > > > > > > Fix is to add 1 to the amount to grow the array. "type - 2 + 1" looks > > a bit inefficient but is a closer match to what is being added and I'd > > like to think javac simplifies the expression. > > > > -phil.
