Add UNO API property tester for read-only properties 'double', 'long',
'short' and 'string'.
Change-Id: I7177e133972058b0272a3cf8493521fbfbcade32
Reviewed-on: https://gerrit.libreoffice.org/52142
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Jens Carl <j.carl43@gmx.de>
Move the header file form the global include path include/test to the
module-specific path test/inc.
Change-Id: I72c1f3b420bc387f4f5b9ab75f30f0e8462c1b75
Reviewed-on: https://gerrit.libreoffice.org/51682
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Jens Carl <j.carl43@gmx.de>
Create UNO API property testers for common, so it's possible to share/use
them between all the UNO API service property tests.
Testers are available for the types 'boolean', 'long', 'short', and
'string'.
Change-Id: Ia36dba98c774695b3ba6ac9d9917af9fb20efdba
Reviewed-on: https://gerrit.libreoffice.org/51259
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Jens Carl <j.carl43@gmx.de>
There are occasional failures of Jenkins "Daily Screenshot Build on Windows"
like <https://ci.libreoffice.org/job/lo_tb_master_win_screenshot/324/console>
and <https://ci.libreoffice.org/job/lo_tb_master_win_screenshot/332/console>,
always failing with
> C:/cygwin/home/tdf/lode/jenkins/workspace/lo_tb_master_win_screenshot/test/source/screenshot_test.cxx(84) : error : Assertion
> Test name: ReportdesignDialogsTest::openAnyDialog
> assertion failed
> - Expression: aNew.IsOpen()
> - Failed to open <C:/cygwin/home/tdf/lode/jenkins/workspace/lo_tb_master_win_screenshot/workdir/screenshots/modules/dbreport/ui/condformatdialog/CondFormat.png>: 796
(where 796 is ERRCODE_IO_NOTEXISTSPATH). Beats me how that can happen, when
ScreenshotTest::implSaveScreenshot first creates the dir and then the file-in-
dir (and it is rather unlikely that something else deletes the dir in the
meantime, for various builds of that Jenkins bot, always for the same dir
workdir/screenshots/modules/dbreport/ui/condformatdialog/), and why it always
happens for exactly the same file,
workdir/screenshots/modules/dbreport/ui/condformatdialog/CondFormat.png during
CppunitTest_reportdesign_dialogs_test.
However, one curiosity was that the dir's URL was computed with
m_directories.getURLFromWorkdir(aDirname);
(without a leading slash inserted before aDirname) while the file's pathname was
computed with
m_directories.getPathFromWorkdir("/" + aDirname + ...)
(with a leading slash inserted before aDirname). Turns out that SvFileStream
accepts a URL as well as a pathname, so normalize the computation of the dir's
and the file's URL. Maybe that will give a clue why that Jenkins bot sometimes
fails.
Change-Id: I53b59b51ffc4355c45aa0ca72f6e187cf2010f92
...regression introduced with eff7034719
"tdf#45904 Move _XFunctionDescriptions Java test to C++" (where
java.util.Random.nextInt(n) returned a value 0 <= x < n), apparently causing
sporadic failures of CppunitTest_sc_functionlistobj like <https://
ci.libreoffice.org/job/lo_gerrit/27696/Config=linux_clang_dbgutil_64/console>
> ##Failure Location unknown## : Error
> Test name: sc_apitest::ScFunctionListObj::testGetById
> An uncaught exception of type com.sun.star.lang.IndexOutOfBoundsException
Also verify that nCount is not negative, so that the requirements of the
std::uniform_int_distribution ctor are guaranteed to be satisfied.
Change-Id: I90a5dc234fdd07f52fea69ae1d406f0818efb007
...that fails for me on at least one macOS and Windows machine each (with 180
and 230, resp., instead of 116), and where it is unclear why only a value of 116
would be correct (and where that specific check was not present before
95307f2ca7 "tdf#45904 Move
_SpreadsheetViewSettings Java tests to C++", cf. my comment from 2018-01-17 at
<https://gerrit.libreoffice.org/#/c/47625/> "tdf#45904 Move
_SpreadsheetViewSettings Java tests to C++")
Change-Id: I16475ecaa853220ffe9e1d6dcf403302b04f4572
Reviewed-on: https://gerrit.libreoffice.org/48215
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
auto-rewrite with <https://gerrit.libreoffice.org/#/c/47798/> "Enable
loplugin:cstylecast for some more cases" plus
solenv/clang-format/reformat-formatted-files
Change-Id: Ia0835f0d9fe59bf0cb42210e4a6be1f022259a8d
As far as I see the old qadevOOo/runner/lib/MultiPropertyTest.java
didn't assert the content of the property value (only avoided the set of
the property if the value was null or void), the new C++ version asserts
both the type and the value. Asserting the type is probably a good
thing, but asserting the value is problematic, at least the test failed
for me without these tweaks.
Change-Id: I688486025e6118ad4c8907de6cb2629dd008f8fd
Reviewed-on: https://gerrit.libreoffice.org/47475
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Jens Carl