Bug 89492 - First Slide Blank in Slide Show
Summary: First Slide Blank in Slide Show
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Impress (show other bugs)
(earliest affected) release
Hardware: Other Windows (All)
: medium normal
Assignee: Not Assigned
Whiteboard: target:4.5.0 4.4.2
Keywords: regression
Depends on:
Reported: 2015-02-20 07:32 UTC by Joey Reid
Modified: 2015-12-22 01:33 UTC (History)
6 users (show)

See Also:
Crash report or crash signature:

Presentation illustrating the issue with blank first slide in 4.5 (11.48 KB, application/vnd.oasis.opendocument.presentation)
2015-02-20 07:32 UTC, Joey Reid

Note You need to log in before you can comment on or make changes to this bug.
Description Joey Reid 2015-02-20 07:32:34 UTC
Created attachment 113539 [details]
Presentation illustrating the issue with blank first slide in 4.5

Steps to Reproduce:
1. File->New-Presentation in a recent build of LO 4.5
2. Add some text in the title and/or body
3. F5

Expected results:
Display text from step 2 

Actual results:
Blank, White screen

Subsequent slides are displayed correctly in slide show mode. 

Tested on Win 7
Build ID: 1d9bfd6e289a1cc8872038b54b060a282d02ea08
Comment 1 Robert Zuccherato 2015-02-21 16:05:32 UTC
I have exactly the same problem except I am running LibreOffice, Build ID: de093506bcdc5fafd9023ee680b8c60e3e0645d7 on Windows 7.
Comment 2 Robert Zuccherato 2015-02-21 22:25:03 UTC
After some experiments, this is what I came up with.  I uninstalled and installed  The problem did not exist.  I could view Impress presentations, including the first slide with no problems.  Therefore, the bug seems to have been introduced in version

I then re-installed version and the bug reappeared.  I messed around with some of the settings and discovered that if I went to Tools->Options->LibreOffice->View and unchecked the "Use hardware acceleration" box then the bug disappears.  All slides can be viewed as normal, including the first slide.

Thus, it seems to be a problem, introduced in with the hardware acceleration feature.  I hope that helps solve the bug and gives people a viable workaround.
Comment 3 Shinji Enoki 2015-02-22 10:30:56 UTC
I was reproduced in the following environments.

LibO: 4.4.0 , 4.4.1 RC1(TDF build)
OS: Windows7

I do not reproduce in Debian.
Comment 4 A (Andy) 2015-02-22 11:07:05 UTC
Reproducible with LO, Win 8.1.  After I deactivated the Hardware Acceleration it also worked for me again.
Comment 5 Tomofumi Yagi 2015-02-26 11:53:42 UTC
I notice that this bug will not reproduced by reverting the following commit. 

"vcl: remove SalVirtualDevice::GetWidth"
commit  c662b1ea318a02747197005ac6547b92d191aab3 


Tested on Windows 8.1
Comment 6 Tomofumi Yagi 2015-02-26 12:17:02 UTC
Probably, we still need GetSize() method calling(please see the following). 

----------  ----------  ---------- 
diff --git a/vcl/source/gdi/virdev.cxx b/vcl/source/gdi/virdev.cxx
index d783cd9..063bef1 100644 (file)
--- a/vcl/source/gdi/virdev.cxx
+++ b/vcl/source/gdi/virdev.cxx
@@ -164,12 +164,6 @@ void VirtualDevice::ImplInitVirDev( const OutputDevice* pOutDev,
     mbScreenComp    = true;
     mnAlphaDepth    = -1;
-    // #i59315# init vdev size from system object, when passed a
-    // SystemGraphicsData. Otherwise, output size will always
-    // incorrectly stay at (1,1)
-    if( pData && mpVirDev )
-        mpVirDev->GetSize(mnOutWidth,mnOutHeight);
     if( mnBitCount < 8 )
         SetAntialiasing( ANTIALIASING_DISABLE_TEXT );
----------  ----------  ----------
Comment 7 Joey Reid 2015-02-26 23:06:56 UTC
Thank you for tracking down the cause of this regression! How were you able to narrow down the regression window without bibisect for Windows? 

Adding Michael Meeks and Markus Mohrhard to make them aware of this issue.

Thanks everyone!
Comment 8 Massimo 2015-02-27 09:30:06 UTC
Same problem here, since version 4.3. I am now at, the bug is still there. Windows 8.1
Comment 9 Massimo 2015-02-27 09:46:41 UTC
(In reply to Massimo from comment #8)
> Same problem here, since version 4.3. I am now at, the bug is still
> there. Windows 8.1

Edit: I confirm that unchecking "Use hardware acceleration" in Tools-> Options-> LibreOffice-> View fixes the problem.
Comment 10 Tomofumi Yagi 2015-02-28 05:48:39 UTC
(In reply to Joey Reid from comment #7)
> @Tomofumi
> Thank you for tracking down the cause of this regression! How were you able
> to narrow down the regression window without bibisect for Windows? 

Thank you.
I used "git bisect", and perseverance.
Because there is no other choice for me :-<
Comment 11 Michael Meeks 2015-02-28 11:36:33 UTC
Really great work; thanks ! =) Unfortunately, that patch is rather useful ;-) it un-screws-up some particularly tortured life-cycle around VirtualDevices that is inexecusably bad =) will need to think this through.

Ultimately, similar sizing logic should happen in the factory / constructor I think:

        mpVirDev = pSVData->mpDefInst->CreateVirtualDevice( pGraphics, nDX, nDY, nBitCount, pData );

To avoid this unpleasant habit of re-sizing virtual-devices later in this unusual way.
Comment 12 Michael Meeks 2015-02-28 12:27:54 UTC
I have a fix vs. master in:


I can't even build-test OS/X which is a bit nerve-racking; hopefully Norbert's CI magic will do the job =) as/when/if that completes - I'll push to master and back-port to 4.4.

I get my first slide on windows, no problems with this - and yet we retain the (relatively saner) semantics - though (to be sure) we're shoving basically 2x use-cases through this API: that with a valid pData and that without - which is a bit lame & prolly error prone.
Comment 13 Michael Meeks 2015-02-28 14:27:24 UTC
Pushed to master, and awaiting review for -4-4 as:


Second opinions appreciated; Markus ? =)
Comment 14 Tomofumi Yagi 2015-02-28 16:08:21 UTC
(In reply to Michael Meeks from comment #12)
> I have a fix vs. master in:
> https://gerrit.libreoffice.org/14678

Thank you very much for your ultra fast fix!

I confirmed that my first slide was displayed correctly on windows 8.1, 
LibreOffice ID: 45306431759ec5e1bda8caa1ed428fc77035b0f2).
Comment 16 Robinson Tryon (qubit) 2015-12-22 01:33:17 UTC
Removing comma from Whiteboard (please use a space to delimit values in this field)