Re: [gccsdk] gcc Digest, Vol 90, Issue 5

Sent from Samsung tablet

-------- Original message --------
From gcc-request@gccsdk.riscos.info
Date: 06/04/2014 21:00 (GMT+01:00)
To gcc@gccsdk.riscos.info
Subject gcc Digest, Vol 90, Issue 5


Send gcc mailing list submissions to
gcc@gccsdk.riscos.info

To subscribe or unsubscribe via the World Wide Web, visit
http://www.riscos.info/mailman/listinfo/gcc
or, via email, send a message with subject or body 'help' to
gcc-request@gccsdk.riscos.info

You can reach the person managing the list at
gcc-owner@gccsdk.riscos.info

When replying, please edit your Subject line so it is more specific
than "Re: Contents of gcc digest..."


Today's Topics:

   1. Re: Diffutils no longer compiles (John Tytgat)


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

Message: 1
Date: Sat, 05 Apr 2014 21:01:28 +0100
From: John Tytgat <John.Tytgat@aaug.net>
To: gcc@gccsdk.riscos.info
Subject: Re: [gccsdk] Diffutils no longer compiles
Message-ID: <8357bff353.Jo@hobbes.bass-software.com>
Content-Type: text/plain; charset=iso-8859-1

In message <DUB120-DS147F359571A432C002CD1DF06C0@phx.gbl>
          "Alan Buckley" <alan_baa@hotmail.com> wrote:

> I???ve just tried compiling the latest diffutils with the GCCSDK
> version 4.1.2 and 4.7 are both the same and it comes up with
> the error:
>
>   CC       gettime.o
> In file included from ./sys/time.h:30,
>                  from gettime.c:24:
> /home/riscos/cross/lib/gcc/arm-unknown-riscos/4.1.2/../../../../arm-unknown-riscos/include/sys/time.h:75: error: conflicting types for 'rpl_gettimeofday'
> ./sys/time.h:401: error: previous declaration of 'rpl_gettimeofday' was here
> make[2]: *** [gettime.o] Error 1
> make[2]: Leaving directory `/home/alanb/build/diffutils/diffutils-3.3/lib'
> make[1]: *** [all] Error 2
> make[1]: Leaving directory `/home/alanb/build/diffutils/diffutils-3.3/lib'
> make: *** [all-recursive] Error 1
>
> I???ve tried to untangle what???s going on and it seems like <sys/time.h> is
> including the local sys/time.h rather than the system one. But I may
> be wrong?
>
> Can someone have a look at it and see how to fix it please?

It's a bit nasty.  Basically there are a lot of configure tests being
done in order to figure out if the used runtime routines adhere to the
standard(s).  A couple of those tests concern gettimeofday().  The last
one wants to check if gettimeofday() and localtime() share a static
buffer (apparently a MacOS X issue).  This test can only be done by
running a small test program but as we're cross-compiling the result is
'guessing yes' instead.  Which means that a replacement 'rpl_gettimeofday'
code piece is being used instead of the system gettimeofday() routine.
And that code gets compiled with an error on systems where the second
parameter type is non-void.

All very messy.  I've worked around by basically enforcing the last
test to 'no' (UnixLib's gettimeofday() and localtime() do not share
a static buffer) with r6638 so diffutils can now be built.

Will you upload a new diffutils port ?

John.
--
John Tytgat, in his comfy chair at home
John.Tytgat@aaug.net



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

_______________________________________________
gcc mailing list
gcc@gccsdk.riscos.info
http://www.riscos.info/mailman/listinfo/gcc

End of gcc Digest, Vol 90, Issue 5
**********************************