Fwd: CppunitTest_sc_shapetest failing after cd966aac6e

[Regina Henschel]

Hi,

That test has
     if (!IsDefaultDPI())
         return;
to skip it for other UI scaling.

So it might be something different? Do you have execute the steps of the 
test in the GUI of running LibreOffice. Is the measure line there 
correct after loading? The difference in the test is about 2cm and 
should be visible.

For me the question is, whether the test is poor or whether the fix does 
not work in all environments. I cannot reproduce the problems on my PC, 
that makes it difficult to investigate.

Kind regards
Regina

Mike Kaganski schrieb am 26.01.2021 um 19:21:
> Missed to CC the list somehow.
> 
> -------- Forwarded Message --------
> Subject: Re: CppunitTest_sc_shapetest failing after cd966aac6e
> Date: Tue, 26 Jan 2021 21:20:38 +0300
> From: Mike Kaganski <mikekaganski at hotmail.com>
> To: Luke Benes <lukebenes at hotmail.com>
> 
> On 26.01.2021 21:16, Luke Benes wrote:
>> After https://cgit.freedesktop.org/libreoffice/core/commit/?id=cd966aac6e
>>      tdf#137033 improve save of cell anchored shapes
>>
>> CppunitTest_sc_shapetest fails with the errors below.  I can reliable 
>> reproduce the failure and the verified the bisect with a 'make clean' 
>> between builds.
>>
>> I am surprised that the build bots missed this. Are all windows bots 
>> doing a 'make check' now that it is required?
> 
> IIRC you use some UI scaling; and that already allowed to find similar 
> bugs that were missed by buildbots that carefully do their make checks, 
> but with default scaling.
> 
>