[gentoo-commits] repo/gentoo:master commit in: dev-util/ruff/

["Sam James" <sam@gentoo.org>]

commit:     f1376d3cec40d14442c4c48acb053ea9a4e3d37c
Author:     Eli Schwartz <eschwartz93 <AT> gmail <DOT> com>
AuthorDate: Fri May 10 01:49:57 2024 +0000
Commit:     Sam James <sam <AT> gentoo <DOT> org>
CommitDate: Fri May 10 01:57:40 2024 +0000
URL:        https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=f1376d3c

dev-util/ruff: fix tests rebuilding jemalloc against vendored copy

I will never understand why people INSIST on "local -x GLOBAL_VAR" as
though it is somehow clean when its fundamental intention is to be
global.

End result: repeating the same block of code multiple times so you can
pseudo-globally set it locally each time you use it. *After* the code is
buggy for a while.

I have a recentish custom QA check that catches it though:

 * QA Notice: cargo built package contains possibly vendored code
 *
 * crate tikv-jemalloc-sys: jemalloc.sym.o arena.sym.o background_thread.sym.o base.sym.o bin.sym.o bin_info.sym.o bitmap.sym.o buf_writer.sym.o cache_bin.sym.o ckh.sym.o counter.sym.o ctl.sym.o decay.sym.o ecache.sym.o div.sym.o edata.sym.o edata_cache.sym.o ehooks.sym.o emap.sym.o eset.sym.o exp_grow.sym.o extent.sym.o extent_dss.sym.o extent_mmap.sym.o fxp.sym.o san.sym.o san_bump.sym.o hook.sym.o hpa.sym.o hpa_hooks.sym.o hpdata.sym.o inspect.sym.o large.sym.o log.sym.o malloc_io.sym.o mutex.sym.o nstime.sym.o pa.sym.o pa_extra.sym.o pai.sym.o pac.sym.o pages.sym.o peak_event.sym.o prof.sym.o prof_data.sym.o prof_log.sym.o prof_recent.sym.o prof_stats.sym.o prof_sys.sym.o psset.sym.o rtree.sym.o safety_check.sym.o sc.sym.o sec.sym.o stats.sym.o sz.sym.o tcache.sym.o test_hooks.sym.o thread_event.sym.o ticker.sym.o tsd.sym.o witness.sym.o jemalloc.o arena.o background_thread.o base.o bin.o bin_info.o bitmap.o buf_writer.o cache_bin.o ckh.o counter.o ctl.o decay.o div.o ecache.o ed
 ata.o edata_cache.o ehooks.o emap.o eset.o exp_grow.o extent.o extent_dss.o extent_mmap.o fxp.o san.o san_bump.o hook.o hpa.o hpa_hooks.o hpdata.o inspect.o large.o log.o malloc_io.o mutex.o nstime.o pa.o pa_extra.o pai.o pac.o pages.o peak_event.o prof.o prof_data.o prof_log.o prof_recent.o prof_stats.o prof_sys.o psset.o rtree.o safety_check.o sc.o sec.o stats.o sz.o tcache.o test_hooks.o thread_event.o ticker.o tsd.o witness.o jemalloc.pic.o arena.pic.o background_thread.pic.o base.pic.o bin.pic.o bin_info.pic.o bitmap.pic.o buf_writer.pic.o cache_bin.pic.o ckh.pic.o counter.pic.o ctl.pic.o decay.pic.o div.pic.o ecache.pic.o edata.pic.o edata_cache.pic.o ehooks.pic.o emap.pic.o eset.pic.o exp_grow.pic.o extent.pic.o extent_dss.pic.o extent_mmap.pic.o fxp.pic.o san.pic.o san_bump.pic.o hook.pic.o hpa.pic.o hpa_hooks.pic.o hpdata.pic.o inspect.pic.o large.pic.o log.pic.o malloc_io.pic.o mutex.pic.o nstime.pic.o pa.pic.o pa_extra.pic.o pai.pic.o pac.pic.o pages.pic.o peak_event.pic.
 o prof.pic.o prof_data.pic.o prof_log.pic.o prof_recent.pic.o prof_stats.pic.o prof_sys.pic.o psset.pic.o rtree.pic.o safety_check.pic.o sc.pic.o sec.pic.o stats.pic.o sz.pic.o tcache.pic.o test_hooks.pic.o thread_event.pic.o ticker.pic.o tsd.pic.o witness.pic.o

 * QA Notice: cargo built package contains possibly vendored code from unsolvable crates
 *
 * ring

Only reproducible when building with FEATURES=test, however, which I
don't usually do on syncing @world.

Signed-off-by: Eli Schwartz <eschwartz93 <AT> gmail.com>
Signed-off-by: Sam James <sam <AT> gentoo.org>

 dev-util/ruff/ruff-0.4.4.ebuild | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/dev-util/ruff/ruff-0.4.4.ebuild b/dev-util/ruff/ruff-0.4.4.ebuild
index 437bdd839436..da6cf3e1d708 100644
--- a/dev-util/ruff/ruff-0.4.4.ebuild
+++ b/dev-util/ruff/ruff-0.4.4.ebuild
@@ -427,6 +427,11 @@ src_compile() {
 }
 
 src_test() {
+	# Gentoo bug #927338
+	if use !elibc_musl && use !elibc_Darwin && use !elibc_bionic; then
+		local -x CARGO_FEATURE_UNPREFIXED_MALLOC_ON_SUPPORTED_PLATFORMS=1
+		local -x JEMALLOC_OVERRIDE="${ESYSROOT}/usr/$(get_libdir)"/libjemalloc.so
+	fi
 	cargo_src_test
 }