catch3, branch v3.5.4 Progmasoft Catch3: an independently maintained C++17 test framework with a parallel runner, property checks, snapshots and structured results. v3.5.4 2024-04-10T10:05:46+00:00 Martin Hořeňovský 2024-04-10T10:05:46+00:00 b5373dadca40b7edc8570cf9470b9b1cb1934d40 






Explicitly outline TestRegistry destructor into .cpp file
2024-04-08T11:57:59+00:00

Martin Hořeňovský

2024-04-08T11:57:53+00:00

cd8f97e6c7f6aaf332992f83bf9d7ee7f357f85e

This fixes compilation issue with C++23 mode against libstdc++.

Closes #2852





This fixes compilation issue with C++23 mode against libstdc++.

Closes #2852







Fix & extend tests for comparing const instances of zero lit types
2024-04-08T11:15:35+00:00

Martin Hořeňovský

2024-04-08T11:15:31+00:00

05fb437cbbf60de794a7101c5e6ef8f650755ce4












Fix assertion for const instance of std::ordering
2024-04-08T09:33:43+00:00

Martin Hořeňovský

2024-04-08T09:02:22+00:00

71b11c4e3328090d2ec72268299e3fb9e0ac3f8e

The issue was that `capture_by_value` was meant to be specialized
by plain type, e.g. `capture_by_value<std::weak_ordering>`, but
the TMP in Decomposer did not properly throw away `const` (and
`volatile`) qualifiers, before taking the value of `capture_by_value`.





The issue was that `capture_by_value` was meant to be specialized
by plain type, e.g. `capture_by_value<std::weak_ordering>`, but
the TMP in Decomposer did not properly throw away `const` (and
`volatile`) qualifiers, before taking the value of `capture_by_value`.







Fix preprocessor check for enabling FP reproducibility tests
2024-04-06T18:27:15+00:00

Martin Hořeňovský

2024-04-06T18:19:47+00:00

0a6a2ce88764cc8c01411fee68834571a5b9a0a1

This solves warning about the `uniform_fp_test_params` helper
being unused on M1 mac (or any other platform using GCC and not
using SSE2).

Closes #2845





This solves warning about the `uniform_fp_test_params` helper
being unused on M1 mac (or any other platform using GCC and not
using SSE2).

Closes #2845







Add M1 MacOS workflow
2024-04-06T18:27:14+00:00

Martin Hořeňovský

2024-04-06T17:53:54+00:00

355a6e273b1bc5d50fd8a066af82c579ad021834












Replace last use of std::uniform_int_distribution with our own
2024-04-03T11:28:26+00:00

Martin Hořeňovský

2024-04-02T16:13:15+00:00

bff6e35e2b239217f3940ed52429f94b745adc50

Our implementation should be slightly faster, and has the
advantage of being consistent between platforms. This does not
have immediate user impact, because we currently use random_device
to generate random seed for resampling, but if we decide to change
this in the future, it is one less place to fix.





Our implementation should be slightly faster, and has the
advantage of being consistent between platforms. This does not
have immediate user impact, because we currently use random_device
to generate random seed for resampling, but if we decide to change
this in the future, it is one less place to fix.







Optimize 64x64 extended multiplication implementation
2024-04-03T11:28:25+00:00

Martin Hořeňovský

2024-04-02T16:09:34+00:00

d99eb8bec8de53639540820ee2a6563b88c4ba8c

Now we use intrinsics when possible, and fallback to optimized
implementation in portable C++. The difference is about 4x when
we can use intrinsics and about 2x when we cannot.

This should speed up our Lemire's algorithm implementation nicely.





Now we use intrinsics when possible, and fallback to optimized
implementation in portable C++. The difference is about 4x when
we can use intrinsics and about 2x when we cannot.

This should speed up our Lemire's algorithm implementation nicely.







Use SizedUnsignedType_t to pick UnsignedType for uniform_integer_distribution
2024-04-03T11:27:10+00:00

Martin Hořeňovský

2024-04-03T10:29:51+00:00

f181de9df4c43f188ae932a412af32a0981c9067

The previously used `make_unsigned` approach combined with the overload
set of `extendedMult` caused compilation issues on MacOS platform. By
forcing the selection to be one of `std::uintX_t` types we don't need
to complicate the overload set further.





The previously used `make_unsigned` approach combined with the overload
set of `extendedMult` caused compilation issues on MacOS platform. By
forcing the selection to be one of `std::uintX_t` types we don't need
to complicate the overload set further.







Merge pull request #2850 from jeremy-rifkin/jr/mention-succeed-in-loggingmd
2024-03-30T14:59:43+00:00

Martin Hořeňovský

2024-03-30T14:59:43+00:00

9271083a0493fe24a56efadcef4c3d640c5dbcad

Mention SUCCEED along with FAIL in logging.md




Mention SUCCEED along with FAIL in logging.md