I just built ghdl-0.27 with gcc-4.1.2 which I still had from building
ghdl-0.26 and that worked just fine. Attached is the script that does
it.

Is there anything wrong with ghdl-0.27 being compiled with gcc-4.1.2
rather than gcc-4.2.4?

Thanks,
Dirk

On Tue, Aug 5, 2008 at 8:11 AM, David G. Koontz <[EMAIL PROTECTED]> wrote:
> Jason Pyeron wrote:
>>>> -----Original Message-----
>>>> From: [EMAIL PROTECTED]
>>>> [mailto:[EMAIL PROTECTED] On Behalf Of Dirk Barthel
>>>> Sent: Monday, August 04, 2008 12:52 PM
>>>> To: GHDL discuss list
>>>> Subject: [Ghdl-discuss] GHDL-0.27 for Cygwin
>>>>
>>>> Hello,
>>>>
>>>> I am trying to build ghdl-0.27 for cygwin. I did it for
>>>> ghdl-0.26 without any problems. Here is the procedure I used for
>>>> ghdl-0.26:
>>>> http://home.comcast.net/~bp_labs/software/ghdl/index.htm
>>>>
>>>> I am having trouble building gcc-4.2.4 for the ghdl-0.27 build.
>>>> I am using the following steps for ghdl-0.27:
>>>> wget ftp://ftp.gnu.org/pub/gnu/gcc/gcc-4.2.4/gcc-core-4.2.4.tar.bz2
>>>> wget ftp://ftp.gnu.org/pub/gnu/gcc/gcc-4.2.4/gcc-ada-4.2.4.tar.bz2
>>>> mkdir gcc-4.2.4-build
>>>> cd gcc-4.2.4-build
>>>> ../gcc-4.2.4/configure -prefix=/usr/local/gcc-4.2.4
>>>> -enable-languages=c,ada >& configure.txt make >& make.txt
>>>>
>>>> Make fails with the following error:
>>>> Comparing stages 2 and 3
>>>> warning: ./cc1-checksum.o differs
>>>> warning: ./libgcc/_chkstk.o differs
>>>> Bootstrap comparison failure!
>>>> ./ada/exp_aggr.o differs
>>>> make[2]: *** [compare] Error 1
>>>> make[2]: Leaving directory `/tmp/gcc-4.2.4-build'
>>>> make[1]: *** [stage3-bubble] Error 2
>>>> make[1]: Leaving directory `/tmp/gcc-4.2.4-build'
>>>> make: *** [all] Error 2
>>
>> Confirmed. I am now trying alternate build options and toolchains.
>
> There's a change in exp_aggr.adb in gcc-4.2.4/gcc/ada/ChangeLog:
>
> 2006-02-13  Ed Schonberg  <[EMAIL PROTECTED]>
>
>        * exp_aggr.adb (Build_Array_Aggr_Code): Rename variable
>        "Others_Mbox_Present" to "Others_Box_Present" because the mbox
>        concept does not exist in the Ada RM.
>        (Compatible_Int_Bounds): Determine whether two integer range bounds
>        are of equal length and have the same start and end values.
>        (Is_Int_Range_Bounds): Determine whether a node is an integer range.
>        (Build_Record_Aggr_Code): Perform proper sliding of a nested array
>        aggregate when it is part of an object declaration.
>        (Build_Record_Aggr_Code)  If the aggregate ttype is a derived type
>        that constrains discriminants of its parent, add explicitly the
>        discriminant constraints of the ancestor by retrieving them from the
>        stored_constraint of the parent.
>
> That should match up to gcc-4.0.3  (2006-03-10) or there abouts.  It would
> appear to be a tool chain incompatibility with the source.
>
> For grins I got the 2008 gnat i686 gnat from libre2.adacore.com and tried to
>  build the ada from 4.2.4.   The first thing I ran into with make
> languages=ada was a bad casing for Standard_ASCII NUL in ali.adb, resulting
> in an error.  The 2008 release is I believe 4.1.3 based which doesn't show
> up in the gcc release directory on gnu.org.  It's doing a style check on
> Standard_ASCII, something that was shown as a patch into the gcc tree around
> 20 May 2008, coincidentally  the day after 4.2.4 was released.  The same
> patch appears to have introduced style checking for it, too.
>
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> Ghdl-discuss mailing list
> [email protected]
> https://mail.gna.org/listinfo/ghdl-discuss
>

Attachment: ghdl-0.27-build.csh
Description: C-Shell script

_______________________________________________
Ghdl-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/ghdl-discuss

Reply via email to