Re: [Libreoffice] smoketest fails
On Mon, 2011-07-25 at 20:46 -0400, Terrence Enger wrote: Greetings, I configured with options --disable-mozilla --enable-symbols --enable-debug --enable-dbgutil --enable-crashdump --without-junit and `make` seemed to complete normally, although I did notice warnings going by. Now, `make dev-install` or `build` in smoketestoo_native/ fail in various ways. I have found http://wiki.documentfoundation.org/Development/How_to_debug#Debugging_cppunit_tests, and I am trying with gdb to work slowly toward the problem. Meanwhile, I wonder if I might be the first to try what I used to know in openoffice as a non-product build? Will the smoketest survive a raised assertion? Hmm, it depends. In indeed might be failing on a debug-only assertion. I do run it in --enable-dbgutil mode but not --enable-debug. For the smoketest itself, if it fails the best thing to do is to run it manually, it's just a document with some macros in it and draws a pretty table at the end of what failed and what passed. So, after a failed make dev-install, you still have an install in solver/350/unxlng*/installation run the soffice from there, turn off macro security with options-security-macros-low and manually open the smoketestdoc.sxw in smoketestdoc/unxlng*/bin and click the start button and see what happens. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fails
On Tue, 2011-07-26 at 10:02 +0100, Caolán McNamara wrote: On Mon, 2011-07-25 at 20:46 -0400, Terrence Enger wrote: (snip) Will the smoketest survive a raised assertion? Hmm, it depends. In indeed might be failing on a debug-only assertion. I do run it in --enable-dbgutil mode but not --enable-debug. I have wimped out: reconfigured without --enable-debug and rebuilt the whole thing; then `make dev-install` worked. For the smoketest itself, if it fails the best thing to do is to run it manually, it's just a document with some macros in it and draws a pretty table at the end of what failed and what passed. So, after a failed make dev-install, you still have an install in solver/350/unxlng*/installation run the soffice from there, turn off macro security with options-security-macros-low and manually open the smoketestdoc.sxw in smoketestdoc/unxlng*/bin and click the start button and see what happens. Neither in the previous build nor in the current does the Macro Security button bring up the dialog I expect. Dunno what I did wrong; just hope it doesn't come back to bite me later. Thank you for your help. Terry. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fail, N12_GLOBAL__N_14TestE::test
On Tue, 2011-05-31 at 08:22 +0200, Julien Nabet wrote: Hello, in smoketestoo_native, I've got this : officeconnection.cxx:140:Assertion Test name: N12_GLOBAL__N_14TestE::test setUp() failed - equality assertion failed - Expected: 2 - Actual : 0 So I used export DEBUGCPPUNIT=TRUE to have gdbtrace.log The smoketest is too big and complicated for the tracer to really help. Do a make dev-install, ignore the failures. go to the install set in solver/*/unxlng*/installation, run it manually, turn off (set to low) macro security in tools-macro and open the smoketest doc in smoketestdoc/unxlng*/bin manually and run it, and see what (if any) boxes go red. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fail, N12_GLOBAL__N_14TestE::test
On Tue, 31 May 2011 08:22:46 +0200 Julien Nabet serval2...@yahoo.fr wrote: in smoketestoo_native, I've got this : officeconnection.cxx:140:Assertion Test name: N12_GLOBAL__N_14TestE::test setUp() failed - equality assertion failed - Expected: 2 - Actual : 0 That assert waits one second with the running LO and expects to get a timeout (i.e. it tests the LO process to be running for at least a second without crashing, before starting tests). You are seeing the started instance return too fast. It should be fixed (hopefully for all cases) with: http://cgit.freedesktop.org/libreoffice/testing/commit/?id=c414bb1dc75b3db4fa88874ff5f978e6a0d18d56 which still needs reviews: http://nabble.documentfoundation.org/REVIEW-3-4-REVIEW-3-4-0-use-nologo-to-prevent-oosplash-smoketest-hangs-td3005829.html Best, Bjoern -- https://launchpad.net/~bjoern-michaelsen ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fails
Am Thu, 26 May 2011 13:44:12 +0100 schrieb Caolán McNamara: On Thu, 2011-05-26 at 13:30 +0200, Christian Dywan wrote: Am Thu, 26 May 2011 09:51:45 +0100 schrieb Caolán McNamara: Still broken for me. I did remove smoketestoo_native/unxlngx6.pro but no difference. Presumably 6eae40f7a1dfbd5 in bootstrap was the commit supposed to fix it? That one was to fix the dangling install dir after a make-install 11f7f9c22fba9551adcfce46657e02376962c7bd in ure was basically the bit to fix the above. So apparently the error can occur due to a wrong LD_LIBRARY_PATH which I had defined due to Java not being found for Netscape plugins. I suggest the error message should be clearer about this. Especially because my system LibreOffice install did find Java and did not have any issues with the path, which contributed to my looking in the wrong direction. -- ciao, Christian ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fails
On Wed, 2011-05-25 at 23:08 +0200, Pim Bollen wrote: [Java framework] The JRE specified by the bootstrap variable UNO_JAVA_JFW_JREHOME or UNO_JAVA_JFW_ENV_JREHOME could not be recognized. Check the values and make sure that you use a plug-in library that can recognize that JRE. javaldx failed! ... - Actual : Extensions:services This should (hopefully) be fixed again now. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fails
Am Thu, 26 May 2011 09:51:45 +0100 schrieb Caolán McNamara: On Wed, 2011-05-25 at 23:08 +0200, Pim Bollen wrote: [Java framework] The JRE specified by the bootstrap variable UNO_JAVA_JFW_JREHOME or UNO_JAVA_JFW_ENV_JREHOME could not be recognized. Check the values and make sure that you use a plug-in library that can recognize that JRE. javaldx failed! ... - Actual : Extensions:services This should (hopefully) be fixed again now. Still broken for me. I did remove smoketestoo_native/unxlngx6.pro but no difference. Presumably 6eae40f7a1dfbd5 in bootstrap was the commit supposed to fix it? -- ciao, Christian ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest fails
On Thu, 2011-05-26 at 13:30 +0200, Christian Dywan wrote: Am Thu, 26 May 2011 09:51:45 +0100 schrieb Caolán McNamara: Still broken for me. I did remove smoketestoo_native/unxlngx6.pro but no difference. Presumably 6eae40f7a1dfbd5 in bootstrap was the commit supposed to fix it? That one was to fix the dangling install dir after a make-install 11f7f9c22fba9551adcfce46657e02376962c7bd in ure was basically the bit to fix the above. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest
On Thu, 2010-11-11 at 01:55 +0100, Jan Holesovsky wrote: - I suppose we want to get rid of unpacking of the installation sets for smoketest, and instead use ooinstall directly ;-) This used to work - I wrote the patch myself ;-) having said that it used (AFAIR) the same pass fifteen magic arguments to make_installer.pl that was used in ooinstall - so, a tad fragile. So yes; I don't think we should use EPM, it should be disabled by default (it takes time), and should use ooinstall for the smoketest ;-) HTH, Michael. -- michael.me...@novell.com , Pseudo Engineer, itinerant idiot ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest
On Thu, 2010-11-11 at 01:55 +0100, Jan Holesovsky wrote: Hi Norbert, Caolán McNamara píše v Po 08. 11. 2010 v 13:47 +: There is some improvement on the smoketest front It run pass many more tests.. but still do not finish: The smoketest passes every time for me. FWIW # A fatal error has been detected by the Java Runtime Environment: # # SIGSEGV (0xb) at pc=0x7f0a3aab45ce, pid=31378, tid=139682275239680 There's clearly a problem, and we should find out what it is. So... What's your distro and version. Versions of java and versions of gcc and glibc ? I have spent most of today trying to get smoketest to work (initially I had similar problems as you, on SLED11 [with some SP1 updates], x86-64), and here are the observations: - you do not want to do make dev-install before make smoketest You see, this is now somewhat confusing. We need a single way to build LibreOffice because I can't tell if you're talking about the makefile in build or the one in rawbuild where there are two different dev-install targets, and I can see that they seem to do two slightly different things, and there isn't a make smoketest in the inner Makefile of course. What I have been doing (with 100% success rate to date on Fedora 14 FWIW) is to run make in the rawbuild dir, and run build in the smoketest dir. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest
Hi Norbert, Caolán McNamara píše v Po 08. 11. 2010 v 13:47 +: There is some improvement on the smoketest front It run pass many more tests.. but still do not finish: The smoketest passes every time for me. FWIW # A fatal error has been detected by the Java Runtime Environment: # # SIGSEGV (0xb) at pc=0x7f0a3aab45ce, pid=31378, tid=139682275239680 There's clearly a problem, and we should find out what it is. So... What's your distro and version. Versions of java and versions of gcc and glibc ? I have spent most of today trying to get smoketest to work (initially I had similar problems as you, on SLED11 [with some SP1 updates], x86-64), and here are the observations: - you do not want to do make dev-install before make smoketest - something gets terribly confused wrt. the links epm packing (that is necessary for the smoketest itself), and the smoketest does not bootstrap correctly then - I suppose we want to get rid of unpacking of the installation sets for smoketest, and instead use ooinstall directly ;-) - in a change pulled from OOo, there is some suspicious code trying to workaround some lifetime issues - I have reverted it in a local build, does not seem to help much, but I am not sure I did it correctly the first time I checked, so rebuilding again All in all - some improvements, but still, not perfect yet ;-) Regards, Kendy ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest
Hi I also get some Java crashes randomly running regcomp, since OOo 3.3rc2. 3.2 was fine so there has to be a change somewhere. On (2010-11-11 01:55), Jan Holesovsky wrote: Hi Norbert, Caolán McNamara píše v Po 08. 11. 2010 v 13:47 +: There is some improvement on the smoketest front It run pass many more tests.. but still do not finish: The smoketest passes every time for me. FWIW # A fatal error has been detected by the Java Runtime Environment: # # SIGSEGV (0xb) at pc=0x7f0a3aab45ce, pid=31378, tid=139682275239680 There's clearly a problem, and we should find out what it is. So... What's your distro and version. Versions of java and versions of gcc and glibc ? I have spent most of today trying to get smoketest to work (initially I had similar problems as you, on SLED11 [with some SP1 updates], x86-64), and here are the observations: - you do not want to do make dev-install before make smoketest - something gets terribly confused wrt. the links epm packing (that is necessary for the smoketest itself), and the smoketest does not bootstrap correctly then - I suppose we want to get rid of unpacking of the installation sets for smoketest, and instead use ooinstall directly ;-) - in a change pulled from OOo, there is some suspicious code trying to workaround some lifetime issues - I have reverted it in a local build, does not seem to help much, but I am not sure I did it correctly the first time I checked, so rebuilding again All in all - some improvements, but still, not perfect yet ;-) Regards, Kendy ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest
On Sat, 2010-11-06 at 10:13 -0500, Norbert Thiebaud wrote: There is some improvement on the smoketest front It run pass many more tests.. but still do not finish: The smoketest passes every time for me. FWIW # A fatal error has been detected by the Java Runtime Environment: # # SIGSEGV (0xb) at pc=0x7f0a3aab45ce, pid=31378, tid=139682275239680 There's clearly a problem, and we should find out what it is. So... What's your distro and version. Versions of java and versions of gcc and glibc ? As an aside, is accessibility enabled or disabled in your desktop ? C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest segfault
On Wed, 2010-10-27 at 17:07 -0500, Norbert Thiebaud wrote: /lo/ref/rawbuild/solver/330/unxlngx6.pro/installation/opt/program/soffice: Isn't really useful to just get it crashed, so... in solenv/inc/settings.mk add gdb --args after AUGMENT_LIBRARY_PATH_LOCAL in the CPPUNITTESTER line run build in smoketestoo_native and type run into gdb when it appears and do thread apply all bt when it crashes If I had to guess I imagine the crash will be in shutdown code, seeing as that's the most fragile part of OOo. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest segfault
On Tue, Oct 26, 2010 at 4:39 AM, Caolán McNamara caol...@redhat.com wrote:
On Mon, 2010-10-25 at 15:39 -0500, Norbert Thiebaud wrote:
On Mon, Oct 25, 2010 at 3:13 PM, Caolán McNamara caol...@redhat.com wrote:
On Sun, 2010-10-24 at 22:03 -0500, Norbert Thiebaud wrote:
I have a segfault running smoketest.
Its cold comfort, but smoketest works for me ok.
can you post the ./bin/g log of where you are at (and mybe your
autogen parameters),
--enable-maintainer-mode --enable-maintainer-mode
--with-distro=LibreOfficeLinux --disable-kde --disable-kde4 --with-git
--with-num-cpus=8 --with-max-jobs=8 --enable-symbols --disable-gnome-vfs
--with-system-cairo --no-create --no-recursion
As a thought, does it make any difference to e.g.
export SAL_USE_VCLPLUGIN=gen before running build in the
smoketestoo_native dir and/or export SAL_USE_VCLPLUGIN=svp (for headless
mode). From your bt I wouldn't think so, but worth perturbing things to
see what might fall out.
n...@tpa10 /lo/ref/rawbuild/smoketestoo_native (master)$ export
SAL_USE_VCLPLUGIN=svp
n...@tpa10 /lo/ref/rawbuild/smoketestoo_native (master)$ build
build -- version: 275224
=
Building module smoketestoo_native
=
Entering /lo/ref/rawbuild/smoketestoo_native/com/sun/star/comp/smoketest
updating: META-INF/MANIFEST.MF (deflated 11%)
updating: com_sun_star_comp_smoketest/META-INF/ (stored 0%)
updating: com_sun_star_comp_smoketest/META-INF/MANIFEST.MF (deflated 11%)
cp -f ../../../../../unxlngx6.pro/class/TestExtension.jar
../../../../../unxlngx6.pro/misc/com_sun_star_comp_smoketest/TestExtension.jar
Making:TestExtension.oxt
updating: META-INF/ (stored 0%)
updating: META-INF/manifest.xml (deflated 51%)
updating: TestExtension.jar (deflated 37%)
updating: TestExtension.rdb (deflated 90%)
Making:com_sun_star_comp_smoketest.dpz
zipdep -- version: 1.12
Multi Platform Enabled Edition
Entering /lo/ref/rawbuild/smoketestoo_native/data
Entering /lo/ref/rawbuild/smoketestoo_native
dmake: Executing shell macro: ls
$(installationtest_instset)/LibO_*_install-arc_$(defaultlangiso).tar.gz
rm -f -r ./unxlngx6.pro/misc/smoketest/user
mkdir -p ./unxlngx6.pro/misc/smoketest/user
:
LD_LIBRARY_PATH=${LD_LIBRARY_PATH+${LD_LIBRARY_PATH}:}/lo/ref/clone/testing/smoketestoo_native/unxlngx6.pro/lib:/lo/ref/rawbuild/solver/330/unxlngx6.pro/lib
/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/cppunittester \
-env:UNO_SERVICES=file:///lo/ref/clone/testing/smoketestoo_native/./unxlngx6.pro/misc/smoketest/services.rdb
\
-env:UNO_TYPES=file:///lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/types.rdb
\
-env:arg-soffice=path:/lo/ref/rawbuild/solver/330/unxlngx6.pro/installation/opt/program/soffice
-env:arg-user=./unxlngx6.pro/misc/smoketest/user \
-env:arg-env=LD_LIBRARY_PATH${LD_LIBRARY_PATH+=$LD_LIBRARY_PATH}
./unxlngx6.pro/lib/libsmoketest.so
-env:arg-doc=./unxlngx6.pro/bin/smoketestdoc.sxw
/lo/ref/rawbuild/solver/330/unxlngx6.pro/installation/opt/program/soffice:
line 154: 15620 Segmentation fault $sd_prog/$sd_binary $@
^CCaught SIGINT. Trying to quit ...
[.]
n...@tpa10 /lo/ref/rawbuild/smoketestoo_native (master)$ export
SAL_USE_VCLPLUGIN=gen
n...@tpa10 /lo/ref/rawbuild/smoketestoo_native (master)$ build
build -- version: 275224
=
Building module smoketestoo_native
=
Entering /lo/ref/rawbuild/smoketestoo_native/com/sun/star/comp/smoketest
:
LD_LIBRARY_PATH=${LD_LIBRARY_PATH+${LD_LIBRARY_PATH}:}/lo/ref/rawbuild/solver/330/unxlngx6.pro/lib
/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/regmerge
../../../../../unxlngx6.pro/misc/com_sun_star_comp_smoketest/TestExtension.rdb
/UCR ../../../../../unxlngx6.pro/misc/TestExtension.urd
:
LD_LIBRARY_PATH=${LD_LIBRARY_PATH+${LD_LIBRARY_PATH}:}/lo/ref/rawbuild/solver/330/unxlngx6.pro/lib
/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/javamaker
-O../../../../../unxlngx6.pro/class -BUCR -nD
-X/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/types.rdb
../../../../../unxlngx6.pro/misc/com_sun_star_comp_smoketest/TestExtension.rdb
/opt/sun-jdk-1.6.0.20/bin/javac -source 1.6 -target 1.6 -classpath
.:../../../../../unxlngx6.pro/class:/opt/sun-jdk-1.6.0.20/jre/lib/rt.jar:.:/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/ridl.jar:/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/jurt.jar:/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/unoil.jar:/lo/ref/rawbuild/solver/330/unxlngx6.pro/bin/juh.jar::/lo/ref/rawbuild/solver/330/unxlngx6.pro/lib
-d ../../../../../unxlngx6.pro/class @/tmp/mk6hIj6f
updating: META-INF/MANIFEST.MF (deflated 11%)
updating: com_sun_star_comp_smoketest/META-INF/ (stored 0%)
updating: com_sun_star_comp_smoketest/META-INF/MANIFEST.MF (deflated 11%)
updating: com/sun/star/comp/smoketest/ (stored 0%)
updating: com/sun/star/comp/smoketest/TestExtension.class (deflated 51%)
updating: com/sun/star/comp/smoketest/TestExtension$_TestExtension.class
(deflated 47%)
cp -f ../../../../../unxlngx6.pro/class/TestExtension.jar
Re: [Libreoffice] smoketest segfault
On Sun, 2010-10-24 at 22:03 -0500, Norbert Thiebaud wrote: I have a segfault running smoketest. Its cold comfort, but smoketest works for me ok. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: [Libreoffice] smoketest segfault
On Mon, Oct 25, 2010 at 3:13 PM, Caolán McNamara caol...@redhat.com wrote: On Sun, 2010-10-24 at 22:03 -0500, Norbert Thiebaud wrote: I have a segfault running smoketest. Its cold comfort, but smoketest works for me ok. can you post the ./bin/g log of where you are at (and mybe your autogen parameters), so I can try to get to the same point and try again (that will hint seriously on whether the failure I see is something screwed-up in my system or not) Norbert C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
