The buildbot builds are all going away (see bug 1236835). The hazard build is orphaned. We need(ed) a new taskcluster-based hazard build. Except that b2g is tier 3 now, so we don't really. But I have this working, it is much simpler, and it will serve as a good base for a taskcluster browser hazard build.

The only issue with this is that it finds a hazard. :( But it's the same paramBuffer false positive that I battle with off and on.

Comment on attachment 8712450 [details] [diff] [review] Implement a taskcluster-based b2g hazard build Review of attachment 8712450 [details] [diff] [review]: ----------------------------------------------------------------- Terrence - can you review the portions of the shell script (build-mulet-haz-linux.sh, build.b2g, and the mulet-hazards mozconfig) that have to do with running the analysis? They're happily quite dull. Greg, all of the job setup is for you. There are some strangenesses I inflicted on shared code, I'll comment on those sections. I suppose I could have split this into two patches with an empty body for build-mulet-haz-linux.sh... ::: testing/taskcluster/scripts/builder/install-packages.sh @@ +8,5 @@ > test -n "$TOOLTOOL_REV" > > tc-vcs checkout $gecko_dir/tooltool $TOOLTOOL_REPO $TOOLTOOL_REPO $TOOLTOOL_REV > > +(cd $gecko_dir; python $gecko_dir/tooltool/tooltool.py --url https://api.pub.build.mozilla.org/tooltool/ -m $gecko_dir/$TOOLTOOL_MANIFEST fetch -c $TOOLTOOL_CACHE) I removed --overwrite because it's deprecated, and I have another patch where I use this to cache tooltool checkouts (not just downloads; that's already cached). It helps with the development debug cycle because the rustc checkout takes quite a while. Probably, though, it'll need to be opt-in since it sticks a file into the checkout. Anyway, for now this is just a removal of a deprecated option.

I split out enabling this on main trees, starting out with just having it on try, until I disabled the original b2g hazard job. That has now been done, though currently it's awaiting the next reconfigure, so it's time to push this portion too.