Re: new snapshot available: coreutils-8.26.57-00381

[Pádraig Brady]

On 02/03/17 20:59, Assaf Gordon wrote:
> Hello Pádraig and all,
> 
>> On Mar 1, 2017, at 22:42, Pádraig Brady <address@hidden> wrote:
>>
>>  https://pixelbeat.org/cu/coreutils-ss.tar.xz
> 
> From a cursory look, I see two failures that might be new:
> 
> First,
> on various Mac OS X, the gnulib test 'test-lock' fails with exit code 134 
> (SIGABRT?)
> (But no coreutils test failures):
> Log here:
>   https://pretest.housegordon.org/g/4461/logs/test-suite.0.log?inlined=1
> I think there was a recent change with it, here:
>   https://lists.gnu.org/archive/html/bug-gnulib/2017-01/msg00046.html

Forwarded to gnulib

> Second,
> On Debian 7.6, 'cut-huge-range' fails, log here:
>  https://pretest.housegordon.org/g/4479/logs/tests-suite-summary.log?inlined=1
> On newer Debian 8.6 there are no failures.

That looks like an auto ulimit issue hopefully avoided with:

diff --git a/tests/misc/cut-huge-range.sh b/tests/misc/cut-huge-range.sh
index c015ac5..6b3c5b6 100755
--- a/tests/misc/cut-huge-range.sh
+++ b/tests/misc/cut-huge-range.sh
@@ -22,6 +22,7 @@ getlimits_

 vm=$(get_min_ulimit_v_ cut -b1 /dev/null) \
   || skip_ "this shell lacks ulimit support"
+vm=$(($vm + 1000)) # avoid spurious failures

 # sed script to subtract one from the input.
 # Each input line should consist of a positive decimal number.

> 
> 
> Few previously known failures (like 'rpl_malloc' on AIX7,

AIX build failure is probably fixed with:

diff --git a/src/libstdbuf.c b/src/libstdbuf.c
index 2abb0f8..e295cc8 100644
--- a/src/libstdbuf.c
+++ b/src/libstdbuf.c
@@ -20,6 +20,11 @@
 #include <stdio.h>
 #include "system.h"

+/* Deactivate config.h's "rpl_"-prefixed definition of malloc,
+   since we don't link gnulib here, and the replacement isn't
+   needed in this case as we don't malloc(0).  */
+#undef malloc
+
 /* Note currently for glibc (2.3.5) the following call does not change
    the buffer size, and more problematically does not give any indication
    that the new size request was ignored:

> 'wc_width' on OpenBSD 5.9,
> gnulib on musl-libc, and few on NetBSD 7.0) - these were all discussed for 
> previous releases so I won't repeat them.
> 
> 
> 
> Additional strange thing is that on two OpenSolaris systems (SunOS 5.11 x86 
> and sparc),
> it seems the test '/tests/misc/timeout.sh' hangs (the command "timeout .5 sh 
> -c 'sleep 2; echo foo'" does not terminate).
> That could be a problem in my setup. I haven't had a chance to investigate 
> further.
> But so far I don't have test results for any OpenSolaris system.

Well drats. timeout(1) has changed to use sigsuspend()
I'll have a look around for a solaris system...

> Also,
> I'm experimenting with easier reporting for these tests,
> these reports are all available here:
>    https://pretest.housegordon.org/q/coreutils-8.26.57
> Clicking on a system name shows a summary for the build, and direct log 
> downloading is at the bottom
> of the page. Comments are very welcomed.

Very nice indeed.

thanks!!
Pádraig