14:13 < mst__> _rene_, the toolkit unoapi tests are known to be flaky (in some 
               system dependent way) e.g. on the Win@6 tinderbox it always 
               crashes
14:14 < mst__> _rene_, sc.ScAccessible* tests also fail on some systems some of 
               the time

--- a/toolkit/Module_toolkit.mk-old	2012-09-05 14:24:27.129693868 +0200
+++ b/toolkit/Module_toolkit.mk	2012-09-05 14:24:39.521755312 +0200
@@ -28,8 +28,8 @@ $(eval $(call gb_Module_add_targets,toolkit,\
 ifneq ($(OOO_JUNIT_JAR),)
 $(eval $(call gb_Module_add_subsequentcheck_targets,toolkit,\
     JunitTest_toolkit_complex \
-    JunitTest_toolkit_unoapi \
 ))
 endif
+#    JunitTest_toolkit_unoapi \
 
 # vim: set noet sw=4 ts=4:
diff --git a/sc/qa/unoapi/sc.sce b/sc/qa/unoapi/sc.sce
index 4ececd9..5eda25a 100644
--- a/sc/qa/unoapi/sc.sce
+++ b/sc/qa/unoapi/sc.sce
@@ -16,21 +16,6 @@
 #   the License at http://www.apache.org/licenses/LICENSE-2.0 .
 #
 # i84653 -o sc.ScPageObj
-# i84554 -o sc.AccessibleEditableTextPara_PreviewNote
-# i88241 -o sc.AccessibleEditableTextPara_HeaderFooter
--o sc.AccessibleEditableTextPara_PreviewCell
--o sc.ScAccessibleCell
-# i91044 -o sc.ScAccessibleCsvCell
-# i91044 -o sc.ScAccessibleCsvGrid
-# i84641 -o sc.ScAccessibleCsvRuler
-# i88330 -o sc.ScAccessibleDocument
-# i91045 -o sc.ScAccessibleDocumentPagePreview
--o sc.ScAccessiblePageHeader
-# fdo#FOO -o sc.ScAccessiblePageHeaderArea
--o sc.ScAccessiblePreviewCell
--o sc.ScAccessiblePreviewHeaderCell
--o sc.ScAccessiblePreviewTable
-# fdo#45337 -o sc.ScAccessibleSpreadsheet
 # FIXME_REMOVE_WHEN_RE_BASE_COMPLETE
 # -o sc.ScAnnotationObj
 # -o sc.ScAnnotationShapeObj

08:54 <@_rene_> vmiklos: 
http://zyklop.dyndns.org/~rene/libreoffice_4.2.4~rc1~git20140414-1_amd64.build.sw_complex. -4-2 from yesterday evening
08:55 <@_rene_> vmiklos: tl;dr: sw_complex fails
[...]
09:25 <@vmiklos> _rene_: i would recommend disabling the checkBookmarks part in 
                 your builds, it's clearly not stable enough to invoke it from 
                 a packaging script.
09:26 <@_rene_> vmiklos: ok, ooc, why? fails sometimes(tm) even for you?
09:26 < tml> sberg: I hate to say this, but maybe we need to take a step 
             backwards and introduce some macro to initialise OUStrings...?
09:26 < tml> s/need/would need
09:27 <@_rene_> why is it then either not fixed or disabled for everyone? I 
                mean, if I patch it now and it gets fixed somewhen I'd stil 
                have it disabled...
09:27 -!- arranna [~ronkko@a88-114-229-111.elisa-laajakaista.fi] has joined 
          #libreoffice-dev
09:27 < loircbot> LibreOffice (core) sbergman * sd/ (inc/randomnode.hxx 
                  source/ui/unoidl/randomnode.cxx): Fix symbol visibility
09:28 <@vmiklos> _rene_: yes, it fails here as well sometimes. the test 
                 generates random bookmarks.
[...]
09:43 <@vmiklos> i might be wrong, but i don't think the checkBookmarks test 
                 was ever stable. if it was stable on 4-1 for you, you're just 
                 lucky :)

diff --git a/sw/JunitTest_sw_complex.mk b/sw/JunitTest_sw_complex.mk
index e9b344c..ffc94f7 100644
--- a/sw/JunitTest_sw_complex.mk
+++ b/sw/JunitTest_sw_complex.mk
@@ -28,7 +28,6 @@ $(eval $(call gb_JunitTest_add_sourcefiles,sw_complex,\
     sw/qa/complex/accessibility/AccessibleRelationSet \
     sw/qa/complex/checkColor/CheckChangeColor \
     sw/qa/complex/indeterminateState/CheckIndeterminateState \
-    sw/qa/complex/writer/CheckBookmarks \
     sw/qa/complex/writer/CheckCrossReferences \
     sw/qa/complex/writer/CheckFlies \
     sw/qa/complex/writer/CheckIndexedPropertyValues \
@@ -58,11 +57,4 @@ $(eval $(call gb_JunitTest_add_classes,sw_complex,\
     complex.writer.TextPortionEnumerationTest \
 ))
 
-# FIXME has never worked on windows, hashes are different
-ifneq ($(OS),WNT)
-$(eval $(call gb_JunitTest_add_classes,sw_complex,\
-    complex.writer.CheckBookmarks \
-))
-endif
-
 # vim: set noet sw=4 ts=4: