WinDef.DWORD getLow() & getHigh() using incorrect bit mask

[imgx64]

#128 was fixed by 245eb88 in 2012, but then 4218a07 reverted it with no explanation given. So I'm reopening it again.

Old bug description:

These two methods are using 0xFF as the bit mask when they should be using 0xFFFF. Each method returns a WORD, which is a 16-bit value. By using 0xFF only the lower byte is preserved.

New Code:

public WORD getLow() {
    return new WORD(longValue() & 0xFFFF);
}

public WORD getHigh() {
    return new WORD((longValue() >> 16) & 0xFFFF);
}

Old Code:

public WORD getLow() {
    return new WORD(longValue() & 0xFF);
}

public WORD getHigh() {
    return new WORD((longValue() >> 16) & 0xFF);
}

[dblock]

It would be useful to track down why/where it was reverted, otherwise needs a new PR and a test.

[imgx64]

The "where" is 4218a07 , the "why" is probably by accident. The commit changed the whitespace on the whole file, which drowned the fact these two lines got reverted.

[dblock]

cc: @wolftobias

[matthiasblaesing]

I created a PR (#683) that reverts the problem and introduces a test, that should prevent breakage in the future. The failing unittest on mac os x fails building the native part, so that result is ignored, unittests on windows work correctly.

If no veto is raised, the PR will be merged sometime next week.