On 2016/01/04 13:57, Christian Weisgerber wrote: > I haven't reported on amd64 bulk build problems in a while, so here > are the ones from the build started on 2016-01-02: > > www/chromium > Just about every build, chromium fails due to some build race. > Or another build race. Or another one. robert@ keeps fixing > them, but there's always another one. I'm getting really tired > of this.
Should we just move this back to gmake? Ninja isn't saving us much time on this port overall, considering the manual restarts that are needed. > x11/vlc > Frequently crashes during fake. sthen@ keeps reporting on this. This seems to happen more on amd64 than i386. IIRC Kili had better luck with the LD_PRELOAD hack back in, though it isn't enough to properly fix it. Maybe it will improve things if we look through LD_DEBUG output and find some other libs which are loaded, unloaded, reloaded several times and preload them as well. > net/mldonkey > "Error: Unbound module Printf2" while compiling > src/daemon/common/commonMain.ml. I don't remember seeing this > before, may not happen again. > > devel/go-check-v1 > devel/go-tools > net/go-websocket > textproc/go-xlsx > Time to mark these broken? Delete them? Does anybody care? I'd go for BROKEN for now. The go-* ports will need bumping to ensure that installed packages get updated to a version built with go 1.5 anyway.