Re: [platform-testers] new snapshot available: coreutils-8.26.61-10d1d

[Assaf Gordon]

Hello,

On Fri, Mar 03, 2017 at 09:25:50AM -0800, Pádraig Brady wrote:
> You can download the coreutils snapshot in xz format (5.2 MB) from:
>  https://pixelbeat.org/cu/coreutils-ss.tar.xz
>  ( coreutils-8.26.61-10d1d )

Most of the issues were fixed, few minor issues remain - likely not show 
stoppers.

1.
The Mac OS X issue ('test-lock') is fixed, no test failures.


2.
On AIX7/PowerPC, the compilation succeeds (I think it's the first time for me)
but the build fails due to 'help2man' issue:
===
GEN      man/test.1
help2man: can't get `--help' info from man/test.td/[
Try `--no-discard-stderr' if option outputs to stderr
make: 1254-004 The error code from the last command is 127.
===



3.
On OpenSolaris 5.10 x86/sparc, no test failures.

On OpenSolaris 5.11 x86/sparc, no coreutils failures,
but one gnulib test failed 'test-mbsalign'.
This is already known:
 https://lists.gnu.org/archive/html/coreutils/2015-07/msg00011.html 



4.
On Debian 7.6, 'tests/misc/cut-huge-range' still fails.
Perhaps it's an issue with my tiny qemu virtual machine that has very 
little memory (512MB or less).



5.
On systems with very old 'makeinfo 4.8' (such as all OpenBSDs and 
FreeBSD 9/10) there is a small packaging regression that won't
affect the actual release, but I'll report anyhow:

The current 'coreutils.texi' contains texinfo commands that don't
work in texinfo 4.8 - that is OK because the minimal version is 4.13.

Normally this isn't a problem because the compiled 'coreutils.info'
is pre-built and pacakged and does not need to be regenerated.

However I suspect that in this current snapshot archive you've prepared
you did not start from a 'maintainer-clean' or 'distclean' state.
The result is that 'doc/stamp-vti' is newer than 'doc/version.texi'
and 'make' needed to regenerate the 'info' file.

Then it failed on those system which have old 'makeinfo' like so:
====
make  all-recursive
Making all in po
Making all in .
GEN      doc/constants.texi
MAKEINFO ./doc/coreutils.info
./doc/coreutils.texi:6: warning: unrecognized encoding name `UTF-8'.
./doc/coreutils.texi:6: Unknown command `allowcodebreaks'.
./doc/coreutils.texi:6568: warning: @sc argument all uppercase, thus no 
effect.
makeinfo: Removing output file `./doc/coreutils.info' due to errors; use 
--force to preserve.
*** Error 1 in . (Makefile:11443 './doc/coreutils.info')
*** Error 1 in . (Makefile:11656 'all-recursive')
*** Error 1 in /tmp/coreutils-8.26.61.FAITG4/coreutils-8.26.61-10d1d 
(Makefile:5922 'all')
====

The timestamp issue can be shown with the following:

 $ wget https://ftp.gnu.org/gnu/coreutils/coreutils-8.26.tar.xz
 $ tar -xf coreutils-8.26.tar.xz 

 $ wget https://pixelbeat.org/cu/coreutils-ss.tar.xz
 $ tar -xf coreutils-ss.tar.xz

 # official release - no problem:
 $ stat -c '%y %n' coreutils-8.26/doc/{coreutils.info,version.texi,stamp-vti} | 
sort
 2016-11-30 13:14:29.000000000 -0500 coreutils-8.26/doc/stamp-vti
 2016-11-30 13:14:29.000000000 -0500 coreutils-8.26/doc/version.texi
 2016-11-30 13:14:54.000000000 -0500 coreutils-8.26/doc/coreutils.info

 # the snapshot release - mixed timestamps:
 $ stat -c '%y %n' 
coreutils-8.26.61-10d1d/doc/{coreutils.info,version.texi,stamp-vti} | sort
 2017-03-03 12:06:26.000000000 -0500 coreutils-8.26.61-10d1d/doc/version.texi
 2017-03-03 12:06:51.000000000 -0500 coreutils-8.26.61-10d1d/doc/coreutils.info
 2017-03-03 12:11:42.000000000 -0500 coreutils-8.26.61-10d1d/doc/stamp-vti


Same recently happened recent in 'sed':
https://lists.gnu.org/archive/html/sed-devel/2017-02/msg00009.html

In practice it's not a big issue because when you prepare a
release archive it'll be generated from scratch and all files will have
the correct time stamps.


------------------

The entire test batch results are available here:
 https://pretest.housegordon.org/q/coreutils-8.26.61


regards,
- assaf