Обсуждение: emu buildfarm failure

Buildfarm member emu is failing in some branches with

checking libintl.h usability... yes
checking libintl.h presence... no
configure: WARNING: libintl.h: accepted by the compiler, rejected by the preprocessor!
configure: WARNING: libintl.h: proceeding with the preprocessor's result
checking for libintl.h... no
configure: error: header file <libintl.h> is required for NLS

I think this is probably because of 
     'config_env' => {                      'CFLAGS' => '-I/usr/local/include',                      'CC' => 'gcc',
                'LDFLAGS' => '-L/usr/local/lib/ -lcrypto -liconv'                     },
 

The config.log trace suggests that libintl.h is installed in
/usr/local/include, because the "working" tests succeed when they have
"-I/usr/local/include" while the "failing" tests don't have that.

-I switches should logically go in CPPFLAGS not CFLAGS.  I'm not sure
this explains emu's failure to build 7.4 and 8.0, because I don't know
why 8.1 and HEAD wouldn't fail the same way ... but it looks wrong
anyway.
        regards, tom lane

Tom Lane wrote:
> Buildfarm member emu is failing in some branches with
> 
> checking libintl.h usability... yes
> checking libintl.h presence... no
> configure: WARNING: libintl.h: accepted by the compiler, rejected by the preprocessor!
> configure: WARNING: libintl.h: proceeding with the preprocessor's result
> checking for libintl.h... no
> configure: error: header file <libintl.h> is required for NLS
> 
> I think this is probably because of 
> 
>       'config_env' => {
>                        'CFLAGS' => '-I/usr/local/include',
>                        'CC' => 'gcc',
>                        'LDFLAGS' => '-L/usr/local/lib/ -lcrypto -liconv'
>                       },
> 
> The config.log trace suggests that libintl.h is installed in
> /usr/local/include, because the "working" tests succeed when they have
> "-I/usr/local/include" while the "failing" tests don't have that.
> 
> -I switches should logically go in CPPFLAGS not CFLAGS.  I'm not sure
> this explains emu's failure to build 7.4 and 8.0, because I don't know
> why 8.1 and HEAD wouldn't fail the same way ... but it looks wrong
> anyway.

sorry for that - c&p error on my part when I reenabled emu to do regular
reports again. I actually tested the conf with a -HEAD build but since
it didn't fail I didn't noticed the error immediately.

The problem itself is fixed now and new reports for all affected
branches should be up very soon.


Stefan