In article <200310280218.46056.abuse@spamalicious.com>,
	abuse@spamalicious.com (Charles M. Hannum) writes:
> On Tuesday 28 October 2003 01:24 am, matthew green wrote:
>>    In article <20031027230305.778422DA1D@cvs.netbsd.org>,
>>
>>    	fredb@netbsd.org (Frederick Bruckman) writes:
>>    > Module Name:	src
>>    > Committed By:	fredb
>>    > Date:		Mon Oct 27 23:03:05 UTC 2003
>>    >
>>    > Modified Files:
>>    > 	src/sys/dev/scsipi: sd.c
>>    >
>>    > Log Message:
>>    > Appease gcc3 -Wno-uninitialized (even though blkno is clearly assigned
>>    > in each branch of a conditional).
>>
>>    It's actually not gcc3 that's stupid: I only just now realized "mac68k"
>>    is still usiing gcc 2.95.3. All such assignments should probably be
>> given a second look after we've completely switched to gcc3.
>>
>>
>> mac68k has switched hasn't it?

Yes, I missed that.

> Yes.  The aforementioned patch should be backed out.

The error occurs with gcc3 after all, with a mac68k cross-toolchain built
today, on an i386 host. My faith in gcc3 was misplaced. 

Frederick