Re: Crash on Linux+Clang (NSTimer)

[David Chisnall]

Hi Riccardo,

On 11/11/2020 23:26, Riccardo Mottola wrote:
> 76% tests passed, 44 tests failed out of 186

Thank you for reporting this.  It looks as if there were some 
regressions in clang's metadata generation after 8.  I saw a bunch of 
failures with clang 10, but the clang 12 nightlies seem somewhat more happy.

I've raised a PR to add more clang configurations to the test matrix:

https://github.com/gnustep/libobjc2/pull/185

This now tests on Ubuntu with all combinations of 64-/32-bit, 
Release/Debug builds, clang-8 and the clang nightly.  I will probably 
add the clang stable branch as well and some point, but it is likely to 
have more failures at the moment and I don't really want to disable 
tests for clang 11 and then reenable them for 12 and later.

It looks as if clang 12 (trunk) is failing in the v1 ABI version of the 
property accessor test.  I'll take a look at that one over the holiday 
and see if we can fix it for the 12 release.

David