[illumos-Developer] Review for 166 CR6901979 error in xdr_float.c not fixed

[Jason King]

Hopefully a final, final update:

http://cr.illumos.org/view/8pdvn2g1/

Basically just ditched the libbc stuff and let it fester, and a minor
tweak to the portable code suggested by Gordon.


On Thu, Mar 31, 2011 at 8:44 PM, Jason King <jason.brian.king at gmail.com> wrote:
> Aside from you, Gordon reviewed it previously, but not the latest
> tweaks.  Not sure if that counts, or if he might perhaps be able to
> look again.
>
> On Thu, Mar 31, 2011 at 1:46 AM, Garrett D'Amore <garrett at nexenta.com> wrote:
>>  Please send me a new webrev.
>>
>> I'd like the Reviewed by: lines in your hg export to reflect your actual
>> reviewers, and a copy of your nightly build's mail_msg file.
>>
>>    - Garrett
>>
>> On 03/21/11 08:42 PM, Jason King wrote:
>>>
>>> On Mon, Mar 21, 2011 at 9:52 PM, Garrett D'Amore<garrett at damore.org>
>>>  wrote:
>>>>
>>>> On Mon, 2011-03-21 at 20:34 -0500, Jason King wrote:
>>>>>
>>>>> Ok.. after a few unrelated issues with the sparc build, this tests out
>>>>> correctly (using data generated from s10 sparc) and also includes a
>>>>> few minor cstyle cleanups in the existing code:
>>>>>
>>>>> http://cr.illumos.org/view/x2ewa65v/
>>>>>
>>>>> _______________________________________________
>>>>> Developer mailing list
>>>>> Developer at lists.illumos.org
>>>>> http://lists.illumos.org/m/listinfo/developer
>>>>
>>>> xdr.h: line 34.  I don't know who Jason Knig is, but its awful nice of
>>>> him to donate his code to us. ;-)
>>>
>>> Doh!
>>>
>>>> xdr_float.c: do you want to #ifdef the xdr_float_portable?  I.e. not
>>>> include it if we have a builtin optimized version?
>>>
>>> I'd need to go back and look (since it's been a while since I touched
>>> this), but ISTR that there was a desire for allowing
>>> testing/interoperability, though my feelings are if you're trying to
>>> deal with floating point across platforms, programmator emptor (i.e.
>>> don't build the portable version on optimized platforms).
>>>
>>> Unless someone chimes up in the next day or two, I'll spin a new
>>> webrev ~Wed with it #ifdef'd out.
>>>
>>> I will also note there is the standing issue of x86 vs sparc
>>> representation of +/-NaN, however that exists independent of this
>>> code, and this does not attempt to wade into that nightmare.
>>>
>>>> Otherwise it looks reasonable to me.
>>>>
>>>>        - Garrett
>>>>
>>>>
>>> _______________________________________________
>>> Developer mailing list
>>> Developer at lists.illumos.org
>>> http://lists.illumos.org/m/listinfo/developer
>>
>>
>> _______________________________________________
>> Developer mailing list
>> Developer at lists.illumos.org
>> http://lists.illumos.org/m/listinfo/developer
>>
>