Suppress clang's (optional) -Wimplicit-fallthrough warnings about switch cases that fall through without a break or return statement. The warnings can also be suppressed with a clang-specific C++11 annotation [[clang::fallthrough]] (aka MOZ_FALLTHROUGH macro from bug 1215411). switch (foo) { case 1: // These cases have no code. No fallthrough annotations are needed. case 2: case 3: foo = 4; // This case has code, so a fallthrough annotation is needed: [[clang::fallthrough]]; // or MOZ_FALLTHROUGH; default: return foo; } gfx/harfbuzz/src/hb-ot-shape-fallback.cc:229:5 [-Wimplicit-fallthrough] unannotated fall-through between switch labels gfx/harfbuzz/src/hb-ot-shape-fallback.cc:263:5 [-Wimplicit-fallthrough] unannotated fall-through between switch labels gfx/harfbuzz/src/hb-ot-shape-fallback.cc:283:5 [-Wimplicit-fallthrough] unannotated fall-through between switch labels And suppress MSVC x64 warnings about 32-bit shifts. I reported these MSVC warnings to the upstream harfbuzz repo: https://github.com/behdad/harfbuzz/issues/146 gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc(240) : warning C4334: '<<' : result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?) gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc(250) : warning C4334: '<<' : result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?) gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc(966) : warning C4334: '<<' : result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?) gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc(1518) : warning C4334: '<<' : result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?) gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc(1672) : warning C4334: '<<' : result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?) gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc(1819) : warning C4334: '<<' : result of 32-bit shift implicitly converted to 64 bits (was 64-bit shift intended?).

Also suppress clang and gcc's -Wshadow warnings and gcc's -Wshadow-local and -Wshadow-compatible-local warnings. These -Wshadow* warnings are not enabled by default, but I am slowly fixing them in other bugs. Do you prefer that I fix these warnings upstream instead of suppressing them in mozilla-central? gfx/harfbuzz/src/hb-open-type-private.hh:382:40 [-Wshadow] declaration shadows a field of 'OT::hb_serialize_context_t' gfx/harfbuzz/src/hb-open-type-private.hh:478:31 [-Wshadow] declaration shadows a field of 'OT::hb_serialize_context_t' gfx/harfbuzz/src/hb-ot-layout-gsubgpos-private.hh:361:48 [-Wshadow] declaration shadows a field of 'OT::hb_apply_context_t' gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc:1250:23 [-Wshadow] declaration shadows a local variable gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc:1625:26 [-Wshadow] declaration shadows a local variable gfx/harfbuzz/src/hb-ot-shape-complex-myanmar.cc:459:23 [-Wshadow] declaration shadows a local variable gfx/harfbuzz/src/hb-ot-shape-complex-use.cc:524:23 [-Wshadow] declaration shadows a local variable gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc:1222:20 [-Wshadow-local] shadowed declaration is here gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc:1254:23 [-Wshadow-local] declaration of 'info' shadows a previous local gfx/harfbuzz/src/hb-ot-shape-complex-myanmar.cc:426:20 [-Wshadow-local] shadowed declaration is here gfx/harfbuzz/src/hb-ot-shape-complex-myanmar.cc:458:23 [-Wshadow-local] declaration of 'info' shadows a previous local gfx/harfbuzz/src/hb-ot-shape-complex-use.cc:483:20 [-Wshadow-local] shadowed declaration is here gfx/harfbuzz/src/hb-ot-shape-complex-use.cc:516:23 [-Wshadow-local] declaration of 'info' shadows a previous local gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc:1595:23 [-Wshadow-compatible-local] shadowed declaration is here gfx/harfbuzz/src/hb-ot-shape-complex-indic.cc:1634:26 [-Wshadow-compatible-local] declaration of 'i' shadows a previous local

Comment on attachment 8675444 [details] [diff] [review] 1215894_suppress-harfbuzz-warnings-v2.patch Punting to Jonathan. My feeling is we should get these upstream but I'll defer to his opinion.

Behdad, are these warnings that you're willing to fix upstream, or do we need to just suppress them locally?

Yes, I'm upstreaming these now. Can even get a release out today.

These should all be fixed. I'll roll a release soon, but would be nice if someone can test them.

Comment on attachment 8675444 [details] [diff] [review] 1215894_suppress-harfbuzz-warnings-v2.patch Thanks, Behdad! I can test your changes soon. There is no rush to make a new release for these warnings. :)

This should have been fixed by harfbuzz rolls since October.