Re: Bison testing on Solaris (with FAILURES)

[Akim Demaille]

Hi Stefano,

Le 26 juin 2013 à 12:42, Stefano Lattarini <address@hidden> a écrit :

> On 06/26/2013 09:57 AM, Stefano Lattarini wrote:
>> Hi Akim,
>> 
>> On 06/24/2013 11:15 PM, Stefano Lattarini wrote:
>>> 
>>> P.S. I'll *try* to run the bison testsuite on non-Linux systems
>>> (Solaris 10, NetBSD 5.1 and FreeBSD 8.2), but I don't know if
>>> I'll have time to do so, so don't hold your breath.  Sorry :-(
>>> 
> OK, tested on Solaris 10 with /bin/ksh as $CONFIG_SHELL, gcc 4.7.1
> as C/C++ compilers, and  /usr/ccs/bin/make as $MAKE.  The build
> succeeded, but there were several testsuite failures.  I attach
> the testsuite.log file (compressed).  It seems all the failure
> are due to the same issue:
> 
>    $(builddir)/lib/stdio.h:1025:1: error: 'char* gets(char*)' \
>        conflicts with previous using declaration \
>        'char* std::gets(char*)'

solaris10-testsuite.log.xz
Description: Binary data

Well, I don't know what to do about this :(  I suppose that line
1025 is something like this:

> /* It is very rare that the developer ever has full control of stdin,
>    so any use of gets warrants an unconditional warning; besides, C11
>    removed it.  */
> #undef gets
> #if HAVE_RAW_DECL_GETS
> _GL_WARN_ON_USE (gets, "gets is a security hole - use fgets instead");
> #endif

that's the only occurrence of gets I found in lib/stdio.in.h.


I could disable the C++ tests at test-suite runtime if the C++
compiler is not happy with a simple #include <iostream>.

I could also try to disable the use of gnulib in our test suite.
Actually, since we are trying to check that our generated headers
work properly and are reasonably portable, that would make a lot
of sense.  However I don't remember why the test suite started
to use gnulib (it might even be by accident, I don't know), and I
don't plan to spend time on detaching the test suite from gnulib
right now.

Do you see options I missed?