[Lldb-commits] [PATCH] D90876: [lldb] [test] Improve comment on expr-after-step-after-crash tests

[Pavel Labath via Phabricator via lldb-commits]

labath added a comment.

In D90876#2377544 <https://reviews.llvm.org/D90876#2377544>, @mgorny wrote:

> In D90876#2377541 <https://reviews.llvm.org/D90876#2377541>, @emaste wrote:
>
>> How does Windows fit into this? Other than that Q, LGTM.
>
> @labath, any clue?

I don't know... Since the test is not annotated, I guess it somehow makes this work, but I haven't been able to find the code that makes it happen.

The problem with this test is that the behavior it prescribes is not compatible with applications that handle SIGSEGV themselves. People have complained about that, including on macos (see D89315 <https://reviews.llvm.org/D89315>). If we were able to tell that the application will not actually handle the SEGV (or other signal), then displaying the process as "crashed" would actually be a good signal to the user. But I don't know of a reasonable way to check whether injecting a signal will cause the app to exit. The situation is probably the same on windows, as it also allows apps to handle their own exceptions (unless we're actually checking whether the application will handle it, which I doubt) , but noone probably noticed it yet...


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D90876/new/

https://reviews.llvm.org/D90876