RFR: JDK-8211029 Have a common set of enabled warnings for all native libraries

[Erik Joelsson]

Looks good.

/Erik


On 2018-09-24 01:18, Magnus Ihse Bursie wrote:
> With JDK-8210988, the foundation is in place for a more systematic way 
> of handling warnings across all native libraries (hotspot and the JDK 
> libraries alike).
>
> With this patch, make sure we enable all warnings equally for all 
> libaries. If an individual library triggers a specific warning, 
> disable it in that library.
>
> There was a single warning from clang in awt_Font.c (due to a very 
> broken cast) that was not possible to turn off (unless -Wextra was 
> turned off entirely), so I fixed the code instead.
>
> I have tested that this compiles without warnings on all standard 
> Oracle build platforms/toolchains. On top of that, I've also tested a 
> variety of gcc's: 4.8.5, 5.5.0, 6.4.0 and 7.3.0 (the minor versions 
> for each major versions shipped by Ubuntu). I've also tested clang 5.0 
> on linux, XCode 9.2 on macosx and Solaris Studio 12.6 on solaris. Even 
> with such extensive testing, the nature of this fix which is to add 
> the flags that turn on a broad range of suitable warnings (like 
> -Wall), and then selectively disable individual warnings on individual 
> libraries, there's a certain risk that some platform/toolchain 
> combinations that used to compile without warnings, now start 
> exhibiting warnings. If this should happen, the short-term workaround 
> is to use "configure --disable-warnings-as-errors". The medium term 
> fix is to disable the problematic warning in the library in question, 
> and the long-term solution is (hopefully) to fix the code.
>
> Bug: https://bugs.openjdk.java.net/browse/JDK-8211029
> WebRev: 
> http://cr.openjdk.java.net/~ihse/JDK-8211029-common-set-of-warnings/webrev.01
>
> /Magnus
>