[8u] RFR (XS) 8219636: Windows build failure after JDK-8207070 8u backport

[Andrew Hughes]

On Mon, 25 Feb 2019 at 11:23, Aleksey Shipilev <shade at redhat.com> wrote:
>
> This is 8u-specific bug (build failure):
>   https://bugs.openjdk.java.net/browse/JDK-8219636
>
> Fix:
>
> diff -r 3a37ebd8c506 src/windows/classes/sun/awt/windows/WToolkit.java
> --- a/src/windows/classes/sun/awt/windows/WToolkit.java Mon Jan 21 19:04:19 2019 -0300
> +++ b/src/windows/classes/sun/awt/windows/WToolkit.java Mon Feb 25 12:16:46 2019 +0100
> @@ -43,10 +43,11 @@
>  import javax.swing.text.JTextComponent;
>
>  import sun.awt.AppContext;
>  import sun.awt.AWTAccessor;
>  import sun.awt.AWTAutoShutdown;
> +import sun.awt.DisplayChangedListener;
>  import sun.awt.LightweightFrame;
>  import sun.awt.SunToolkit;
>  import sun.misc.ThreadGroupUtils;
>  import sun.awt.Win32GraphicsDevice;
>  import sun.awt.Win32GraphicsEnvironment;
>
> Testing: {Linux,Windows} x86_64 build
>
> Separately: Paul, have you tested the backport patch builds with 8u on Windows?
>
> Let me propose/reiterate the ground rule for requesters: before doing Fix Request, you need to
> actually apply the patch, test it, make sure it works, and then Fix Request it, mentioning the
> testing you have done and changes you needed to make it work. If the patch touches the OS you don't
> have access to, _do not_ take the backport in the works.
>
> Maintainers have to cross-check that testing is adequate for the backport being proposed. Yes, that
> is more work, but that is why non-rubber-stamping reviews are needed: they routinely catch simple
> mistakes.
>
> -Aleksey
>

Moreover, it should be possible to build the WIndows and Mac OS X Java
classes manually to check
they still build. That's what I do for 6 & 7 [0]

Going forward, it is also the responsibility of those who want to
continue support for these platforms
to ensure the build doesn't break on them. I don't think it's
practical to expect every contributor to
build on GNU/Linux, Windows, Mac OS X and Solaris. I certainly don't
have the means to do so.
Not backporting code that touches platform-specific code is one thing.
Not backporting anything
that might break an untested platform is another.

[0] https://bitbucket.org/gnu_andrew/bin/raw/becd42b57bec0f8f193677e5dde34575b9d5dca5/compile_windows.sh
-- 
Andrew :)

Senior Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Web Site: http://fuseyism.com
Twitter: https://twitter.com/gnu_andrew_java
PGP Key: ed25519/0xCFDA0F9B35964222 (hkp://keys.gnupg.net)
Fingerprint = 5132 579D D154 0ED2 3E04  C5A0 CFDA 0F9B 3596 4222