Mark Wielaard wrote:
On Sun, 2009-07-12 at 15:00 -0400, Orcan Ogetbil wrote:
> On Sun, Jul 12, 2009 at 1:54 PM, Mark Wielaard wrote:
>> I am curious. What was the problem using gcj?
> Check the link above. It gave some 3000+ mysterious compilation
> warnings, compared to 68 using openjdk:
I must be missing something. It does look like those warnings are
genuine. Isn't it just the case that gcj has more warnings (enabled) by
default?
It's just ecj. ecj comes out with millions of warnings.
Basically, it's the compiler saying "I don't like your coding style.
Please try to be a better programmer and I'll shut up." I think the
compiler should just STFU anyway and get on with its job.
The Sun javac equivalent for ecj's default is -Xlint, in which case it
emits nearly as many warnings as ecj. But this only happens if you increase
the maximum warning count; otherwise it just prints the first 100 warnings.
I don't know why IBM chose to make warnings=on the default, unlike Sun
and just about every other compiler provider. But Sun still prints a
few really useful warnings, even without -Xlint.
I don't think there is an ecj equivalent of Sun's default.
Now that we've got a separate ecj RPM we could provide one.
Andrew.