a4a53cdd24
2009-09-01 sb #i76393# second attempt at properly #ifdef-ing previous HG commit d598efdbf012
2009-08-28 sb #i102469# change back <T extends XInterface> to just <T> on queryInterface, to avoid binary incompatibility (method changing its signature from (Ljava/lang/Class;Ljava/lang/Object;)Ljava/lang/Object; to (Ljava/lang/Class;Ljava/lang/Object;)Lcom/sun/star/uno/XInterface;)
2009-08-28 sb #i76393# properly #ifdef previous HG commit d598efdbf012
2009-08-27 sb #i94421# work around compiler error (based on a patch supplied by cloph)
2009-08-26 sb merged in DEV300_m56
2009-08-26 sb #i76393# on Linux, include dynamic section offset in crash report so as to be able to map "prelinked" callstacks back to original (patch by cmc)
2009-08-26 sb #i88162# remove unnecessary whitespace lines from per-locale xcu files (patch by tora)
2009-08-17 Juergen Schmidt #i104292# set context classloader after create new custom UNO loader
2009-08-17 Juergen Schmidt #i103749# integrate patch
2009-08-14 sb #i103269# cherry-picked ssh://hg@hg.services.openoffice.org/cws/sb111 -r 5124ebd5edd1 ("#i101955# changed encoding of XML file content from erroneous ISO-8859-1 to UTF-8")
2009-08-12 sb #i102469# fixed mis-applications of UnoRuntime.queryInterface (detected via the simplified UnoRuntime.queryInterface, the HG changeset 29de35fc9554) to use AnyConverter instead; changed qadevOOo's lib.MultiMethodTest.before to allow throwing arbitrary exceptions, to cater for IllegalArgumentException thrown by AnyConverter
2009-08-12 sb #i104178# drop extra libxml2-config script from libxmlsec
2009-08-10 sb #i101754# simplified osl_getProcessInfo for LINUX (patch by cmc)
2009-08-10 sb #i95018# avoid closing -1 fds (patch supplied by cmc)
2009-08-10 sb #i103585# removed (apparently unnecessary) zlib support from libxml2; in turn, removed zlib dependencies from libxmlsec, libxslt, and redland (assuming those were transitive dependencies brought in by direct dependencies on libxml2)
2009-08-10 sb #i102469# simplified UnoRuntime.queryInterface using Java 5 generics; adapted URE-related modules accordingly
2009-08-10 sb #i101213# adapted setsolar env (solenv/config/) to set PYTHONPATH (and not set PYTHONHOME) in accordance with configure env (set_soenv.in); fixed testtools/source/bridgetest/pyuno (which now should work everywhere out of the box, thanks to the fixed setsolar PYTHONPATH)
2009-08-10 sb cherry-picked ssh://hg@hg.services.openoffice.org/cws/sb111 -r ea8de6d9396b ("#i101955# work in progress for a .hgignore file, continued")
|
||
|---|---|---|
| .. | ||
| com/sun/star/comp/bridge | ||
| inc | ||
| prj | ||
| qa | ||
| source | ||
| util | ||
| readme.txt | ||
How to check compatibility between compilers ============================================ Since the interfaces used in the cpp bridgetest are not changed often one can just build the cppobj.uno.dll and the constructors.uno.dll (testtools/source/bridgetest) in an old environment and then use them in the new environment. That is the files are copied into the testtools/wntmsciXX.pro folder which corresponds to the new environment. On Windows this test will typically fail because the tests use the cppu::getCaughtException function, which only works when all libs are build using the same runtime. This part of the test can switched off. To do this go into the testtools/source/bridgetest folder and call dmake compcheck=1 This will add a new compiler define (-DCOMPCHECK) and will be used in the bridgetest.cxx to switch off the code which uses the getCaughtException function. However, there is still a test which causes the test component to throw and IllegalArgumentException. This still works. Using source/bridgetest for stress testing ========================================== Start a modified bridgetest_server (with the final "--singleaccept" argument removed from the uno executable call) or a modified bridgetest_javaserver (with the final "singleaccept" argument replaced with "multi" in the java executable call), then start a modified bridgetest_client (with a final "stress" argument added to the uno executable call). The client will continuously establish connections to the server which are immediately destroyed again. The test will run forever, unless an error occurs.