[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: coreutils-8.29 released [stable]

From: Michael
Subject: Re: coreutils-8.29 released [stable]
Date: Tue, 2 Jan 2018 11:40:34 +0100
User-agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:56.0) Gecko/20100101 Thunderbird/56.0

On 01/01/2018 18:50, Michael wrote:
On 01/01/2018 14:58, Pádraig Brady wrote:
On 29/12/17 15:46, Michael wrote:
On 27/12/2017 19:47, Pádraig Brady wrote:
This is to announce coreutils-8.29, a stable release.
See the NEWS below for more details.
Just want to add - that building and packaging - no issues (one warning
message only).

However, cannot run "make check" as this fails with:

address@hidden:[/data/prj/gnu/coreutils-8.29]make  V=1 check-TESTS
/bin/sh: not found

make: 1254-004 The error code from the last command is 1.

make: 1254-004 The error code from the last command is 2.


Happy and Healthy 2018!!
Another issue with AIX make?
Note we've documented in README for this release that
gmake is preferred for AIX. Can you try that?

The problem is AIX 5.3 make - gmake brought me further - and gave a better error message:


make[4]: Entering directory '/data/prj/gnu/coreutils-8.29'
make[4]: execvp: /bin/sh: Arg list too long

On AIX 6.1

/usr/bin/make check is progressing normally. Sorry for the 'noise'.

Anyway -back to original subject.

a) AIX 5.3 /bin/sh is not suitable for make check. This is what is needed in your README re: AIX.

b) regarding the info on AIX in README - the first paragraph is not correct, the second can be expanded - slightly.

Currently README has:

OSF/1 4.0d and AIX build failures

If you use /usr/bin/make on these systems, the build will fail due
to the presence of the "[" target.  OSF/1 make(1) appears to
treat "[" as some syntax relating to locks, while AIX make(1)
appears to skip the "[" target.  To work around these issues
the best solution is to use GNU make.  Otherwise, simply remove
all mention of "[$(EXEEXT)" from src/Makefile.

AIX make(1) does not skip the "[" target - at least not since AIX 5.3. Maybe this stems from AIX 4.X days (or earlier).

mkinstallp does not like to package '[' as this character is used as part of the mkinstallp grammar (I have my own workaround that renames '[' before packaging, and creating a link during installp "config" phase. In any case, AIX make(1) has no issue with building '[''


address@hidden:[/data/prj/gnu/coreutils-8.29]find . -name '[' -ls
20582197  137 -rwxr-xr-x  1 199      1954       139619 Jan  1 17:44 ./src/[
address@hidden:[/data/prj/gnu/coreutils-8.29]file ./src/[
./src/[: 64-bit XCOFF executable or object module not stripped

32 bit time_t build failures

On systems where it's determined that 64 bit time_t is supported
(indicated by touch -t <some time after 2038>), but that coreutils
would be built with a narrower time_t, the build will fail.
This can be allowed by passing TIME_T_32_BIT_OK=yes to configure,
or avoided by enabling 64 bit builds.  For example GCC on AIX defaults
to 32 bit, and to enable the 64 bit ABI one can use:
./configure CFLAGS=-maix64 LDFLAGs=-maix64 AR='ar -X64'

I used

# export OBJECT_MODE=64

and build using xlc using the same arguments to configure as for 32-bit.

p.s., even with GCC - I would recommend using export OBJECT_MODE=64 as all IBM-AIX tools will switch to -X64 mode, not just $AR.

c) The test results:

* FAIL: tests/misc/shred-remove
Going to guess here that this is a kind of 'secure' remove process that likely fails because this is an NFS share rather than a local disk. If so, the test needs review, if not - I'll look deeper for you.

Details I found:

+ compare_ exp out
+ diff -u exp out
--- exp 2018-01-01 22:36:09 +0000
+++ out 2018-01-01 22:36:09 +0000
@@ -2,4 +2,5 @@
 shred: test: renamed to 0000
 shred: 0000: renamed to 001
 shred: 001: renamed to 00
+shred: 00: renamed to _
 shred: test: removed
+ fail=1
+ mkdir rodir
+ cd rodir
+ touch 0123456789
+ chmod a-w .
+ returns_ 1 timeout 10 shred -u 0123456789
shred: 0123456789: failed to remove: Permission denied
+ Exit 1
+ set +e
+ exit 1
+ exit 1
+ remove_tmp_
+ __st=1
+ cleanup_
+ :
+ test '' = yes
+ cd /data/prj/gnu/coreutils-8.29
+ chmod -R u+rwx /data/prj/gnu/coreutils-8.29/
+ rm -rf /data/prj/gnu/coreutils-8.29/
+ exit 1
FAIL tests/misc/ (exit status: 1)

* FAIL: tests/cp/preserve-mode


I assume this is what 'fails'

++ cut -b-10
+ test drwxr-sr-x = drwxr-xr-x
+ fail=1


reply via email to

[Prev in Thread] Current Thread [Next in Thread]