Yes, I think this is probably worth fixing.

There's also the related problem that if you specify 

inspector/tracing/ [ Slow ]
inspector/tracing/foo.html [ Failure ]

the [ Slow ] is dropped from foo.html. Some times this is the right thing to do, but most of the time it isn't, and is probably why we have so many flaky timeouts for inspector tests. 

(I've been thinking about this myself recently).

Unfortunately, we don't have a [ Fast ] equivalent to [ Pass ]; we probably either  need to add one or make the propagation rules for Slow work entirely differently than Failure.

Having virtual tests inherit the slowness of the non virtual test suite seems reasonable to me. It's a bit magic, but it's what you want 99% of the time and the consequences of marking too many things as Slow isn't that bad for the remaining 1%.

Naina recently added lint checks that you can only use Slow to the SlowTests file and can't put anything other than Slow in the SlowTests file. The files actually merge their expectations. So, I think the related problem Dirk pointed out is not actually a problem, right?

Long ago when I was working on this code, if a file had a specific expectations line, it would override *any* expectations on enclosing directories (and it had to).

Unless we've specifically added code to *not* do that for Slow tests -- and I don't think we have based on what've seen over the past week -- I think it's still a problem.

I don't think that's how it works if we're merging across files. I might be wrong though.

I think if you have:
SlowTests:
foo [ Slow ]

TestExpectations:
foo/bar.html [ Failure ]

Then foo/bar.html will get [ Slow Failure ].

Hm. At first test, it looks like you're right, but now I'm puzzled by why I'm seeing so many inspector timeouts, then ...

Fixing component name after Monorail migration

Blink>ToolsTest renamed to Blink>Infra

Deprecating component:Blink>LayoutTests, to use label Test=Layout instead.

Hmmm... Alex found out that lack of inheritance appears to be by design according to https://chromium.googlesource.com/chromium/src/+/master/docs/testing/layout_tests.md

I am still not sure if I agree with this design...

FWIW, the problem of lack of inheritance can be seen in https://crrev.com/2505933002 and in https://crrev.com/2547733002

Yes. Feel free to change the implementation if you find that we reduce the total number of expectations by doing so. I don't think it's clear a priori which is better.

This is a P2 but hasn't seen any activity in over a year. Is it really a P2, and if so should someone be assigned? FWIW, seems like a P3 to me.

P3 is fine.