On 22/02/18 22:24, Ythogtha wrote: > Hi everybody! > > I'm almost sure I should have found how to deal with that, because it > feels like I did before, or heard of it here, but I couldn't find anything. > > So I'm trying to build LibreOffice, and it fails at the point described below. > I have removed any previous LibreOffice (or libreoffice, or even openoffice) > package. > It fails wether I set USE_CCACHE=yes or USE_CCACHE=no > It fails whatever value I set to MAKEFLAGS=-j1/-j2/-j4/-j8 > I have openjdk8 installed, and it fails wether I set JAVA=openjdk8 or > JAVA=none. > It fails wether I set VLC=yes or VLC=no. > I have 130Go in the build and tmp directory. > I have 32Go of RAM. > None of my partitions are even near full. > And it fails always at the same place. > > Here is the end of the build output > [......] > [BIN] sc > [MOD] sc > [BIN] postprocess > [GAL] arrows > [GAL] computers > [GAL] diagrams > Bootstrap exception 'component context fails to supply service > com.sun.star.ucb.UniversalContentBroker of type > com.sun.star.ucb.XUniversalContentBroker' > /tmp/fs/SBo/libreoffice-6.0.1.1/solenv/gbuild/Gallery.mk:58: > recipe for target > '/tmp/fs/SBo/libreoffice-6.0.1.1/workdir/Gallery/arrows.done' > failed make[1]: *** > [/tmp/fs/SBo/libreoffice-6.0.1.1/workdir/Gallery/arrows.done] Error 1 make[1]: > *** Waiting for unfinished jobs.... Bootstrap exception 'component context > fails > to supply service com.sun.star.ucb.UniversalContentBroker of type > com.sun.star.ucb.XUniversalContentBroker' > /tmp/fs/SBo/libreoffice-6.0.1.1/solenv/gbuild/Gallery.mk:58: > recipe for target > '/tmp/fs/SBo/libreoffice-6.0.1.1/workdir/Gallery/computers.done' failed > make[1]: > *** [/tmp/fs/SBo/libreoffice-6.0.1.1/workdir/Gallery/computers.done] Error 1 > Makefile:268: recipe for target 'build' failed make: *** [build] Error 2 > > Any insight would be greatly appreciated, thanks ! >
I sometimes saw this type of output in my i586 test builds in a VM, even after allocating more and more resources (RAM, disk space) to it. BTW, no such problems with x86_64 VM, so I've long suspected some sort of memory addressing problem. After running with MAKEFLAGS-=-j1, I noticed a mention about virtual memory exhaustion. This led to more resource allocation increases in the VM, including adding more swap space but still no conclusive improvement (sometimes failure, sometimes success). Eventually, based on an idea which is not quite clear enough in my own mind to try to explain, I set TMP=/var/tmp/SBo when running the SlackBuild (overriding its TMP=/tmp/SBo). Since then the i586 VM has built LO without problems using MAKEFLAGS=-j4 (LO versions 6.0.0.3 & 6.0.1.1). chris _______________________________________________ SlackBuilds-users mailing list SlackBuilds-users@slackbuilds.org https://lists.slackbuilds.org/mailman/listinfo/slackbuilds-users Archives - https://lists.slackbuilds.org/pipermail/slackbuilds-users/ FAQ - https://slackbuilds.org/faq/