https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117342
--- Comment #11 from David Binderman <dcb314 at hotmail dot com> ---
(In reply to Andreas Schwab from comment #10)
> Can you provide an example of the evidence?
Sure. For this C++ source code:
void s61() { static char extra_special_characters[] = "\n\t\b\r\f\\\'"; }
compiled by various gcc trunk C++ compilers over the last few days:
cvise $ for i in /home/dcb40b/gcc/results.2024102*.asan.ubsan/bin/g++; do echo
$i; $i -c cq.cc; done
/home/dcb40b/gcc/results.20241022.asan.ubsan/bin/g++
/home/dcb40b/gcc/results.20241023.asan.ubsan/bin/g++
/tmp/cciIhb4R.s: Assembler messages:
/tmp/cciIhb4R.s:8: Error: unknown pseudo-op: `.base64'
/home/dcb40b/gcc/results.20241024.asan.ubsan/bin/g++
/home/dcb40b/gcc/results.20241025.asan.ubsan/bin/g++
/home/dcb40b/gcc/results.20241028.asan.ubsan/bin/g++
/tmp/cc8b8rIn.s: Assembler messages:
/tmp/cc8b8rIn.s:8: Error: unknown pseudo-op: `.base64'
cvise $
It does look most odd that this works some of the time.
For the git hashes:
cvise $ for i in /home/dcb40b/gcc/results.2024102*.asan.ubsan/bin/g++; do echo
$i; $i -v 2>&1 | grep exp; done
/home/dcb40b/gcc/results.20241022.asan.ubsan/bin/g++
gcc version 15.0.0 20241022 (experimental) (bf11ecbb02b517df)
/home/dcb40b/gcc/results.20241023.asan.ubsan/bin/g++
gcc version 15.0.0 20241023 (experimental) (f342d66d6990b955)
/home/dcb40b/gcc/results.20241024.asan.ubsan/bin/g++
gcc version 15.0.0 20241024 (experimental) (0dbc588acaa27a3a)
/home/dcb40b/gcc/results.20241025.asan.ubsan/bin/g++
gcc version 15.0.0 20241024 (experimental) (17d7a4b01a54b1af)
/home/dcb40b/gcc/results.20241028.asan.ubsan/bin/g++
gcc version 15.0.0 20241028 (experimental) (e320846fec00aaa3)
cvise $
