Commit graph

10 commits

Author SHA1 Message Date
Douglas Katzman 42549b22cc De-flake a test of sxhash hash quality
I asked Gemini to do the math on this, and it said that it's no surprise
a flake was seen with only 20 objects.  Changing to 60 objects reduces the
chance of spurious failure to 1 in 5e-16
2026-09-05 11:14:43 -04:00
Douglas Katzman 9daea02bac win32: disable agressive stack scrub
Some checks are pending
Linux arm / build (push) Waiting to run
CL-host / ecl (push) Waiting to run
CL-host / clisp (push) Waiting to run
CL-host / ccl (push) Waiting to run
CL-host / cmucl (push) Waiting to run
CL-host / sbcl (push) Waiting to run
CL-host / compare-xc-host-fasls (ccl, false) (push) Blocked by required conditions
CL-host / compare-xc-host-fasls (clisp, false) (push) Blocked by required conditions
CL-host / compare-xc-host-fasls (cmucl, false) (push) Blocked by required conditions
CL-host / compare-xc-host-fasls (self, false) (push) Blocked by required conditions
Linux qemu / build (riscv64) (push) Waiting to run
Linux / build (x86, --with-sb-thread, ) (push) Waiting to run
Linux / build (x86, --without-sb-thread, ) (push) Waiting to run
Linux / build (x86, --without-sb-unicode, ) (push) Waiting to run
Linux / build (x86-64, --with-mark-region-gc --with-nonstop-foreign-call) (push) Waiting to run
Linux / build (x86-64, --with-sb-fasteval --without-sb-eval --with-nonstop-foreign-call, fasteval) (push) Waiting to run
Linux / build (x86-64, --with-sb-thread --with-nonstop-foreign-call --with-tls-based-mv-return, sse4) (push) Waiting to run
Mac / build (arm64, --with-sb-thread --with-nonstop-foreign-call) (push) Waiting to run
Mac / build (x86-64, --with-mark-region-gc --with-nonstop-foreign-call) (push) Waiting to run
Mac / build (x86-64, --with-sb-thread --with-nonstop-foreign-call) (push) Waiting to run
Windows arm64 / build (arm64, clang-aarch64, clangarm64) (push) Waiting to run
Windows / build (x86-64, ucrt-x86_64, ucrt64) (push) Waiting to run
Linux arm64 / build () (push) Waiting to run
Linux qemu / build (ppc64le) (push) Waiting to run
Linux / build (x86-64, --with-sb-thread, ) (push) Waiting to run
Linux / build (x86-64, --without-sb-thread, ) (push) Waiting to run
Linux / build (x86-64, --without-sb-unicode, ) (push) Waiting to run
Mac / build (arm64, --with-mark-region-gc --with-nonstop-foreign-call) (push) Waiting to run
2026-09-01 09:59:48 -04:00
Douglas Katzman 310019ac0e Add a hack to make GC weak-hash-table tests reliable
The optimization of not scrubbing the part of the control stack that you
think you don't have to scrub is contrary to the requirement of some tests.
Real use would not often care. As mentioned in the prior commit, the
scrubbing logic was actually deficient going back nearly 15 years,
so the new logic is merely differently deficient but geared toward avoiding
touching all memory of all stacks of all threads on every GC.
2026-09-01 00:04:57 -04:00
Stas Boukarev c5a49f87da Relax a test.
Some checks failed
CL-host / ecl (push) Waiting to run
CL-host / clisp (push) Waiting to run
CL-host / ccl (push) Waiting to run
CL-host / cmucl (push) Waiting to run
CL-host / sbcl (push) Waiting to run
CL-host / compare-xc-host-fasls (ccl, false) (push) Blocked by required conditions
CL-host / compare-xc-host-fasls (clisp, false) (push) Blocked by required conditions
CL-host / compare-xc-host-fasls (cmucl, false) (push) Blocked by required conditions
CL-host / compare-xc-host-fasls (self, false) (push) Blocked by required conditions
Linux / build (x86, --with-sb-thread, ) (push) Waiting to run
Linux / build (x86, --without-sb-thread, ) (push) Waiting to run
Linux / build (x86, --without-sb-unicode, ) (push) Waiting to run
Linux / build (x86-64, --with-mark-region-gc) (push) Waiting to run
Linux / build (x86-64, --with-sb-fasteval --without-sb-eval, fasteval) (push) Waiting to run
Linux / build (x86-64, --with-sb-thread, ) (push) Waiting to run
Linux / build (x86-64, --with-sb-thread, sse4) (push) Waiting to run
Linux / build (x86-64, --without-sb-thread, ) (push) Waiting to run
Linux / build (x86-64, --without-sb-unicode, ) (push) Waiting to run
Mac / build (--without-sb-thread, x86-64) (push) Waiting to run
Mac / build (arm64, --with-mark-region-gc) (push) Waiting to run
Mac / build (arm64, --with-sb-thread) (push) Waiting to run
Mac / build (x86-64, --with-mark-region-gc) (push) Waiting to run
Mac / build (x86-64, --with-sb-thread) (push) Waiting to run
Windows / build (push) Waiting to run
Linux qemu / ppc64le (push) Failing after 0s
2024-11-26 07:07:01 +03:00
Stas Boukarev 4da0692ae1 Disable a test on mark-region-gc.
It either fails, hangs, or prints a warning.
2024-05-20 04:09:41 +03:00
Stas Boukarev 6e6334adf3 Silence failing tests. 2023-06-15 20:28:50 +03:00
Stas Boukarev 79c1bc6c68 Add a test for psxhash on large floats. 2022-07-28 17:12:07 +03:00
Douglas Katzman 780c709169 Unchange test's expectation
Restores externalizability of some hash-table literals with nonstandard test
2022-02-03 11:09:02 -05:00
Douglas Katzman 83ba95f4b7 Change test's expectation 2022-02-03 10:56:24 -05:00
Douglas Katzman 559f9e0049 Enforce fewer side-effects of "pure" regression tests
- Capture before/after state of globaldb and global symbols for comparison.
- Do a better job clobbering remants of defining forms after each test file.
- Allow DEFSTRUCT and DEFTYPE in pure tests.

The intent of this change is to  get a larger set of tests that don't fork
a new child, which will provide more metrics on the effectiveness of GC
during the regression run.

Contrary to the comments, DEFUN and DEFMACRO do not render a test impure,
because DELETE-PACKAGE undoes those. Apparently thread creation is treated
as pure, evidenced by the mere existence of 'threads.pure.lisp'.
Also, creating a file is not impure, because state change outside the process
is usually irrelevant. A failing pure test may have side-effects by accident,
which isn't great, but we can live with it.
2022-01-10 00:54:05 -05:00
Renamed from tests/hash.impure.lisp (Browse further)