liborcus / boost1.83 and time_t
Hi,
I got
qahelper.cxx:580:Assertion
Test name: testContentGnumeric::TestBody
assertion failed
- Expression:
xServiceInfo->supportsService("com.sun.star.sheet.SpreadsheetDocument")
Failures !!!
Run: 64 Failure total: 1 Failures: 1 Errors: 0
make[3]: *** [/<<PKGBUILDDIR>>/solenv/gbuild/CppunitTest.mk:130:
/<<PKGBUILDDIR>>/workdir/CppunitTest/sc_subsequent_filters_test.test]
Error 1
make[3]: Leaving directory '/<<PKGBUILDDIR>>'
make[2]: *** [Makefile:277: build] Error 2
make[2]: Leaving directory '/<<PKGBUILDDIR>>'
make[1]: *** [/<<PKGBUILDDIR>>/debian/rules:2501: check] Error 2
make[1]: Leaving directory '/<<PKGBUILDDIR>>'
make: *** [debian/rules:2389: debian/stampdir/build-arch] Error 2
on the libreoffice 4:24.2.2-1 build on the buildds after it is now
buildable again due to being held up by time_t.
The gnumeric filter mentioned here is consumed from liborcus.
liborcus hasn't been rebuilt yet since the boost1.83-default transition.
It also doesn't depend on libboost-chrono1.83 which is afaics the boost
library which was renamed for time_t. (just iostreams - and for whatever
reason also system on sparc64).
And iostreams doesn't depend on chrono...
boost though has been rebuilt for time_t. And liborcus does use
time-date parts of boost (which doesn't result in linkage)
Back before the transition started without anything actually rebuilt yet
I got a test failure in a test build with abi=+time64 unless I rebuilt
xmlsec1. Which is reflected in the build-depends.
Anyways, as this gnumeric filter is consumed from liborcus I thought
about this being a fallout of time_t and tried to rebuild liborcus.
Rebuilding liborcus and dpkg -i'ing it and then running the test makes
the test pass where it failed here before, too.
That would mean a bin-NMU of liborcus would work and then a rebuild of
libreoffice (gb, but I need a new upload anyway)
So we probably missed a rename? (Or more for stuff silently using
time-date?)
boost1.83 (iostream)? liborcus? Both?
Regards,
Rene
Reply to: