There's a couple of different issues here, maybe?

First is whether the `sizes` step detected the new SI and failed (in a functional sense). It looks like it didn't, right?

Second is whether the static initializer *also* produced a real perf regression, in the sense that some perf benchmark should've moved. Do we know whether the problem is that we're not running a sensitive enough benchmark? Is it possible that SIs just don't produce a big enough impact?

About the `sizes` step not failing: that was filed in  bug 832854  and duped into  bug 832854 .

There aren't specific performance benchmarks for static initializers. The chromium commit position for the change is 558576, so you can go to http://chromeperf.appspot.com/group_report?rev=558576 to look for performance regressions with the CL in range. It doesn't look like a regression was found. If there is a measurable performance regression from this CL, we could talk about adding a benchmark at chrome-benchmarking-request@, but I can't find evidence of one.

@sullivan - Your first paragraph of #c2 confuses me.  Bug 832854  was a case where `sizes` *did* fail, not a case where it didn't. So I don't think this is a dup?

Ah okay, too many problems with the static initializers in sizes to keep up with.

Downgrading P2 issues unmodified for more than 6 months, which have no owner or component.