Build failure on NetBSD at RELEASE_4_5_0-16-g5cba8ae

[James E Keenan]

On 8/3/12 2:26 PM, Andy Dougherty wrote:
> On Wed, 18 Jul 2012, Andy Dougherty wrote:
>
>> On Tue, 17 Jul 2012, Jonathan "Duke" Leto wrote:
>>
>>> Howdy,
>>>
>>> It looks like Parrot_get_cpu_type is only defined in
>>> src/platform/generic/misc.c and src/platform/win32/misc.c but not
>>> src/platform/netbsd/misc.c .
>>>
>>> Not sure if that is the issue, but it seems likely.
>>
>> Yes, that would explain it.  The "correct" way to have done it, within
>> parrot's src/platform hierarchy, would have been to make two new files,
>> something like generic/cpu_type.c and win32/cpu_type.c, along with the
>> appropriate tweaks to MANIFEST and the root Makefile.  Then no special
>> NetBSD tweaks would be needed.
>
> The attached patch implements this.  I tested it on Linux and NetBSD.  I
> have not tested it on Windows.
>
>

I tested this patch my usual 5 ways tonight (4x on Linux/i386, 1x on 
Darwin/PPC).  I had to tweak the MANIFEST a bit, but then applied it to 
master.

I didn't test it on Win32, to which I have no current access.  But I 
figured that, at this point in the monthly release cycle, we're more 
likely to get failures reports from our vast array of Win32 smokers than 
by waiting around for tests of the patch.  And Andy usually gets these 
things right.

Thank you very much.
Jim Keenan