Pádraig Brady <[email protected]> writes:

> On 16/04/2026 19:19, Collin Funk wrote:
>> Hi Bruno,
>> Bruno Haible via GNU coreutils General Discussion
>> <[email protected]>
>> writes:
>> 
>>> On Alpine Linux 3.18 (32-bit x86), there are two test failures:
>>>
>>> FAIL: tests/misc/io-errors
>>> FAIL: tests/cksum/b2sum
>>>
>>> The cksum/b2sum failure looks much like the one that I've reported for
>>> Slackware 15.
>> Thanks for all the testing!
>> Pádraig, do you have any issues with the change I proposed earlier
>> to
>> deal with musl's strange definition of "fully buffered" [1]. If not, I
>> can push it with the adjustments to use $host_os instead of
>> $host_triple. That should fix the misc/io-errors test.
>
> Yes please push that.

I see you pushed it. I pushed the attached follow up patch to replace
the rest of the uses $host_triplet with $host_os.

>> I'm guessing the cksum/b2sum failure is due to a different type limit on
>> 32-bit on x86, but I haven't looked into it much.
>
> I've pushed a fix for that.

Thanks!

Collin

>From 70293149b5aea287f1728579243fadc67d38bbe7 Mon Sep 17 00:00:00 2001
Message-ID: <70293149b5aea287f1728579243fadc67d38bbe7.1776400152.git.collin.fu...@gmail.com>
From: Collin Funk <[email protected]>
Date: Thu, 16 Apr 2026 21:07:26 -0700
Subject: [PATCH] tests: Avoid accidental matching of the vendor field of
 $host, part 2

* tests/local.mk (TESTS_ENVIRONMENT): Remove host_triplet.
* tests/chgrp/basic.sh: Use $host_os instead of $host_triplet.
* tests/misc/io-errors.sh: Likewise.
---
 tests/chgrp/basic.sh    | 2 +-
 tests/local.mk          | 1 -
 tests/misc/io-errors.sh | 2 +-
 3 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/tests/chgrp/basic.sh b/tests/chgrp/basic.sh
index 1d8715efe..4a1dabe62 100755
--- a/tests/chgrp/basic.sh
+++ b/tests/chgrp/basic.sh
@@ -102,7 +102,7 @@ test "$(ls -C -c -t f g)" = 'f  g' || \
       openbsd*) echo ignoring known OpenBSD-specific chgrp failure 1>&2 ;;
       darwin7.9.*|darwin8.*)
         echo ignoring known MacOS X-specific chgrp failure 1>&2 ;;
-      *) echo $host_triplet: no-change chgrp failed to update ctime 1>&2;
+      *) echo $host_os: no-change chgrp failed to update ctime 1>&2;
             fail=1 ;;
     esac
   }
diff --git a/tests/local.mk b/tests/local.mk
index 74b77f220..628e54436 100644
--- a/tests/local.mk
+++ b/tests/local.mk
@@ -68,7 +68,6 @@ TESTS_ENVIRONMENT =				\
   built_programs='$(built_programs) $(single_binary_progs)' \
   fail=0					\
   host_os=$(host_os)				\
-  host_triplet='$(host_triplet)'		\
   srcdir='$(srcdir)'				\
   top_srcdir='$(top_srcdir)'			\
   CONFIG_HEADER='$(abs_top_builddir)/$(CONFIG_INCLUDE)' \
diff --git a/tests/misc/io-errors.sh b/tests/misc/io-errors.sh
index 72bfeb458..75f50050f 100755
--- a/tests/misc/io-errors.sh
+++ b/tests/misc/io-errors.sh
@@ -88,7 +88,7 @@ while read writer; do
   # As a result, when we print a single line there is no bytes buffered when
   # we close the stream and errno is not set.  See:
   # <https://www.openwall.com/lists/musl/2026/04/02/1>.
-  case $host_triplet in
+  case $host_os in
     *-musl*) error_re="$error_re|" ;;
   esac
 
-- 
2.53.0

Reply via email to