When I noticed the warning, actually I tried the exactly same fix (both
equals and hashCode). It was supposed to generate the same data after
the fix, but the fixed one didn't generate identical data. I haven't had
time to look into it further.
Yesterday Yuka looked into the fixed tool and confirmed that the fix
didn't affect the runtime. She found the equals part of the fix actually
fixed the tool behavior (redundant loop cycles, etc.).
So I think the fix should be pushed.
Masayoshi
On 6/6/2013 5:30 AM, Naoto Sato wrote:
I'd compare the generated break iterator data binary before and after
the fix to make sure it is not affecting the compatibility.
Naoto
On 6/5/13 1:27 PM, Alan Bateman wrote:
On 05/06/2013 18:47, Naoto Sato wrote:
What about the regression risk? With this fix, the equality would
differ from the prior releases. Would that be ignorable?
Naoto
This is the tool used in the build, it's the generated data files are
that included in the runtime. I've run the regression tests on all
platforms, is there anything else that I should run?
-Alan.
