--- Comment #22 from Glenn Adams <gadams@...> 2012-04-25 23:27:38 UTC ---
(In reply to comment #20)
> (In reply to comment #19)
> > Please also see comment 13. I think some modifications to the patch are
> > necessary.
> I don't know why Jeremias asked for two parameters for number of decimal places
> (one general, and one for [-1,+1] range). Maybe he planned to use this patch
> also for "transformation matrices" with values in range [-1,+1], but that's
> just a guess.
> Anyway, I believe that we could use this patch without problems since it really
> does not change current behavior of the trunk. The trunk will always call
> format method with those two parameters being equal: (3,3), (4,4), (8,8), etc.
> If you think that there should be change in the way FOP handles doubles, in
> terms of always using 8 decimal places, is it possible to open separate issue?
> I would really like to close this one, and potentially solve memory leak
> problems introduced by ThreadLocal.
i'm more interested in solving the immediate problem than changing the way FOP
handles doubles; i agree it could be a separate issue (if there is a desire to
address it... but i don't know there is a need)