Re: netbsd-7 i386 floating point troubles

[christos%astron.com@localhost (Christos Zoulas)]

In article <smu1s3nzc2w.fsf%linuxpal.mit.edu@localhost>,
Greg Troxel  <gdt%lexort.com@localhost> wrote:
>Greg Troxel <gdt%lexort.com@localhost> writes:
>
>> While testing an update to proj 6 (geospatial coordinate conversions), I
>> noticed that the tests failed on NetBSD 7 i386, after passing on NetBSD
>> 8 amd64.
>>
>> I then ran paranoia (pkgsrc/benchmarks/paranoia), which reported
>> multiple serious problems:
>>
>>   The number of  FAILUREs  encountered =       3.
>>   The number of  SERIOUS DEFECTs  discovered = 4.
>>   The number of  DEFECTs  discovered =         3.
>>   The number of  FLAWs  discovered =           2.
>>
>>   The arithmetic diagnosed has unacceptable Serious Defects.
>>   Potentially fatal FAILURE may have spoiled this program's subsequent
>diagnoses.
>>   
>
>With CFLAGS=-ffloat-store I find htat parnaoia has "1 FLAW" -- but
>that's vastly better -- on both of my i386 machines.
>
>Does anybody understand if compiling without float-store is supposed to
>be ok?

I believe it is not supposed to be ok.

christos