distro diversity in autobuilds

[Stefan Metzmacher]

Am 04.05.2018 um 06:47 schrieb Douglas Bagnall via samba-technical:
> On 22/03/18 15:37, Martin Schwenke via samba-technical wrote:
>> On Thu, 22 Mar 2018 13:43:47 +1300, Douglas Bagnall via samba-technical
>> <samba-technical at lists.samba.org> wrote:
>>
>>> On 22/03/18 12:07, Douglas Bagnall via samba-technical wrote:
>>>> From time to time, Andreas says something like:  
>>>>> as I'm probably the only one who actually runs our tests on a
>>>>> different distro than Ubuntu, [...]  
>>>>
>>>> Which, along with recent Python 2.6 regressions, has prompted me to
>>>> try running something else on the Catalyst cloud. And after much
>>>> tweaking, we can successfully run... {drum-roll}... Debian!
>>>> (Debian 9/stretch/stable to be precise).
>>>>
>>>> But I also have plans for Centos 6.6 and 7.
>>>>
>>>> In Centos 6.6, I arrive here in samba.stderr:
>>>>
>>>> [ 671/4143] Compiling lib/util/become_daemon.c
>>>> cc1: warnings being treated as errors
>>>> ../lib/util/become_daemon.c: In function ‘daemon_ready’:
>>>> ../lib/util/become_daemon.c:128: error: declaration of ‘daemon’ shadows
>>>> a global declaration
>>>> /usr/include/unistd.h:956: error: shadowed declaration is here
>>>>
>>>> GCC is "gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-18)".
>>>>
>>>> Commit a50092531af8c837c6e45c3c755fcb20b6ce323b looks culpable.  
>>>
>>> So I fixed this one (attached) but now it fails on a similar thing a
>>> moment later:
>>>
>>> [ 701/4151] Compiling lib/util/tini.c
>>> cc1: warnings being treated as errors
>>> ../lib/util/tfork.c: In function ‘tfork_status’:
>>> ../lib/util/tfork.c:809: error: declaration of ‘wait’ shadows a global declaration
>>> /usr/include/sys/wait.h:116: error: shadowed declaration is here
>>> Waf: Leaving directory `/home/centos/autobuild/b16891/samba/bin'
>>> Build failed:  -> task failed (err #1): 
>>> 	{task: cc tfork.c -> tfork_32.o}
>>> make: *** [all] Error 1
>>>
>>> which suggests I should just find a way of turning off this silly
>>> warning. Therefore I NACK my own patch and recommend you do the same
>>> unless you think gcc 4.4.7 was onto something important.
>>
>> I broke it, the compiler is stupid, we should fix it...
>>
>> Reviewed-by: Martin Schwenke <martin at meltin.net>
>>
>> Provided, if course, you nix your own NACK.  ;-)
> 
> Nup, I have decided my NACK stands.
> 
> Your 'daemon' here and Ralph's 'wait' in lib/util/tfork.c are
> exemplary names. Changing them for an obsolete compiler would be
> silly.
> 
> What I have tried instead is the attached patch which turns off
> -Werror=shadow for compilers with this problem. But I'm not proposing
> that either, because it turns out not to be sufficient.

I'd propose to add an option to autobuild.py to avoid using
--picky-developer in order to support older systems. Similar to the
AUTOBUILD_NO_EXTRA_PYTHON check. Maybe AUTOBUILD_NO_PICKY_DEVELOPER=1

metze

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20180504/01ddb0fc/signature.sig>