diff mbox

std::polar requires non-negative rho

Message ID 20150513133233.GH30202@redhat.com
State New
Headers show

Commit Message

Jonathan Wakely May 13, 2015, 1:32 p.m. UTC 
http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4383.html#2459

Voted into the WP in Lenexa.

Tested powerpc64le-linux, comitted to trunk.

Comments

Daniel Krügler May 13, 2015, 1:36 p.m. UTC | #1 
2015-05-13 15:32 GMT+02:00 Jonathan Wakely <jwakely@redhat.com>:
> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4383.html#2459
>
> Voted into the WP in Lenexa.
>
> Tested powerpc64le-linux, comitted to trunk.

What about assertions regarding non-NAN rho and finite theta, as
decided for by the LWG 2439?

- Daniel
Jonathan Wakely May 13, 2015, 2 p.m. UTC | #2 
On 13/05/15 15:36 +0200, Daniel Krügler wrote:
>2015-05-13 15:32 GMT+02:00 Jonathan Wakely <jwakely@redhat.com>:
>> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4383.html#2459
>>
>> Voted into the WP in Lenexa.
>>
>> Tested powerpc64le-linux, comitted to trunk.
>
>What about assertions regarding non-NAN rho and finite theta, as
>decided for by the LWG 2439?

I don't know how to reliably test those conditions. What if _Tp is not
a floating-point type?

The assertion I added addresses the original issue as reported
numerous times to libc++. I don't really care if it doesn't deal with
the additional preconditions added in Cologne, as users don't seem to
expect a meaningful answer for NaNs and infinities.
Marc Glisse May 13, 2015, 2:05 p.m. UTC | #3 
On Wed, 13 May 2015, Daniel Krügler wrote:

> 2015-05-13 15:32 GMT+02:00 Jonathan Wakely <jwakely@redhat.com>:
>> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2015/n4383.html#2459
>>
>> Voted into the WP in Lenexa.
>>
>> Tested powerpc64le-linux, comitted to trunk.
>
> What about assertions regarding non-NAN rho and finite theta, as
> decided for by the LWG 2439?

non-NAN rho is already covered by rho >= 0.
Daniel Krügler May 13, 2015, 5:08 p.m. UTC | #4 
2015-05-13 16:05 GMT+02:00 Marc Glisse <marc.glisse@inria.fr>:
> On Wed, 13 May 2015, Daniel Krügler wrote:
>> What about assertions regarding non-NAN rho and finite theta, as
>> decided for by the LWG 2439?
>
> non-NAN rho is already covered by rho >= 0.

Agreed on that.

- Daniel
diff mbox

Patch

commit 9bf3b9ea20334711ecdced656323f69959521a82
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Wed May 13 14:18:03 2015 +0100

    	* include/std/complex (polar): Check for negative rho (LWG 2459).

diff --git a/libstdc++-v3/include/std/complex b/libstdc++-v3/include/std/complex
index 585683c..f2a6cf9 100644
--- a/libstdc++-v3/include/std/complex
+++ b/libstdc++-v3/include/std/complex
@@ -667,7 +667,10 @@  _GLIBCXX_BEGIN_NAMESPACE_VERSION
   template<typename _Tp>
     inline complex<_Tp>
     polar(const _Tp& __rho, const _Tp& __theta)
-    { return complex<_Tp>(__rho * cos(__theta), __rho * sin(__theta)); }
+    {
+      _GLIBCXX_DEBUG_ASSERT( __rho >= 0 );
+      return complex<_Tp>(__rho * cos(__theta), __rho * sin(__theta));
+    }
 
   template<typename _Tp>
     inline complex<_Tp>