Re: PATCH: More protected symbol tests

[Roland McGrath <roland at redhat dot com>]

> I'm going to revert this.  It is stupid to put failing tests in.  One
> should be able to remember enabling the tests once the issue is resolved
> one way or the other.  But disrupting the testing for no real reason is
> really bad.  

When there is a known bug, I think it is better to be reminded of it by the
test rather than letting the sources suggest we don't know it exists.  The
existing test fails to test half of what it claims to test, and it is easy
to overlook an obscure problem like this if it goes for a few days without
resolution and then other things come up.  Having the test for it in the
tree makes everyone aware something needs to be addressed before the next
release.

> People should not run the tests with make -k.  This just incerases the
> possibility to overlook a real bug.

Bullshit.  People who run make check should pay attention to the output
and note all failures.  It's their business how they use make.