Re: CppunitTest_sw_layoutwriter failing on Windows with HiDPI

On 20/09/2019 13:51, Stephan Bergmann wrote:
> On 20/09/2019 08:54, Stephan Bergmann wrote:
> > I experience the same, but on a non-scaled (to the best of my 
> > knowledge) Windows 8 box.  I initially came up with 
> > <https://gerrit.libreoffice.org/#/c/79065/> "Fix 
> > CppunitTest_sw_uiwriter for my Windows build", but see Miklos' comment 
> > there.  I haven't yet found out why it behaves differently than it 
> > apparently should on my Windows box.
>
> So we pushed that Gerrit change as-is for now, to get builds unstuck.

The problem apparently is the call to

>             ::SwAlignRect( aRect, static_cast<SwViewShell const *>(m_pCursorShell), 
> m_pCursorShell->GetOut() );

in SwVisibleCursor::SetPosAndShow (sw/source/core/crsr/viscrs.cxx), 
which modifies aRect (SwAlignRect in sw/source/core/layout/paintfrm.cxx 
says: "adjust rectangle on pixel level in order to make sure, that the 
border 'leaves its original pixel', if it has to").  On my Linux box 
that transforms from 1x277@(1473,1750) to EMPTY:0x276@(1475,1750), while 
on my Windows box it transforms from 1x277@(6287,1418) to 
EMPTY:0x271@(6290,1420).

Miklos, in light of that maybe you have an idea how to improve your test?