Support for different compilers
Magnus Ihse Bursie
magnus.ihse.bursie at oracle.com
Mon Feb 3 19:25:27 UTC 2014
On 2014-02-03 16:17, Volker Simonis wrote:
> As always, you forget the Windows/IA64 which can be only cross-compiled:)
> But OK, I'm pretty sure we won't support Java 9 on Windows/IA64 and I
> hope Java 8 neither.
> Theoretically you could also cross-compile Windows/AMD64 on a 32-bit
> Windows box, but that's probably a more exotic case nowadays where
> everybody has a 64-bit Windows machine.
I'm trying hard to forget IA64. :-) But you are of course correct. But
then again, cross-compiling on Windows is slightly different. Or rather,
compiling to the same platform with a different address length is
different than on other platforms, since it's actually more of a real
cross-compilation. In theory, we could support a proper
cross-compilation from a 32-bit Windows build machine to a Windows
64-bit, but we decided it was too much effort for an unrealistic
scenario. The "reduced" build is not, like in other toolchains, achieved
with a flag, but with a different compiler, so it's sort-of
cross-compilation, but with the difference that the resulting binary can
be run on the build platform.
>> The remaining
>> two are fixes for old brokenness in Solaris tools. With some luck, they are
>> not needed anymore if we can require a new enough OS version/toolchain.
>> Then there's a remaining bad guy, the adlc compiler in Hotspot.
>> Unfortunately, this is a large tool, and converting it to Java would be a
>> more massive undertaking. :-(
> Yes, I agree. I don't see this happening any time soon also that would be nice.
I think we can start trying to get rid of the easy one's, and make sure
we don't add any new, even though we can't get rid of adlc in a while. I
think it still makes sense trying to use the build-cc as little as possible.
More information about the build-dev