Remove a static initializer from ScopedCAPIHandle

Remove a static initializer from ScopedCAPIHandle

BUG=87171
TEST=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=98923

[Ryan Sleevi, 2011-08-31 01:08:21]

Evan: Rubber stamp to mirror your scoped_ptr<> changes?

Backstory on why this isn't wrapping scoped_ptr<> is that it's more akin to
ScopedHandle-et-al (base/win/scoped_handle.h), in that the CAPI types are opaque
ULONG_PTRs, not actual pointer types.

However, given that each type is a ULONG_PTR, but with a unique free function,
the templated approach ala scoped_ptr<> was used, rather than individual and
complete types like ScopedHandle, ScopedHDC, ScopedHGlobal...

[Evan Martin, 2011-08-31 02:16:55]

LGTM

On Tue, Aug 30, 2011 at 6:08 PM,  <rsleevi@chromium.org> wrote:
> Reviewers: Evan Martin,
>
> Message:
> Evan: Rubber stamp to mirror your scoped_ptr<> changes?
>
> Backstory on why this isn't wrapping scoped_ptr<> is that it's more akin to
> ScopedHandle-et-al (base/win/scoped_handle.h), in that the CAPI types are
> opaque
> ULONG_PTRs, not actual pointer types.
>
> However, given that each type is a ULONG_PTR, but with a unique free
> function,
> the templated approach ala scoped_ptr<> was used, rather than individual and
> complete types like ScopedHandle, ScopedHDC, ScopedHGlobal...
>
> Description:
> Remove a static initializer from ScopedCAPIHandle
>
> BUG=87171
> TEST=none
>
>
> Please review this at http://codereview.chromium.org/7776025/
>
> SVN Base: svn://svn.chromium.org/chrome/trunk/src
>
> Affected files:
>  M crypto/scoped_capi_types.h
>
>
> Index: crypto/scoped_capi_types.h
> diff --git a/crypto/scoped_capi_types.h b/crypto/scoped_capi_types.h
> index
>
f565121059da687b3d5c3ac532435cfc9603aafd..51e2f72eb6c37f0f38e31968b4163369f9420085
> 100644
> --- a/crypto/scoped_capi_types.h
> +++ b/crypto/scoped_capi_types.h
> @@ -51,12 +51,13 @@ class ScopedCAPIHandle {
>   explicit ScopedCAPIHandle(CAPIHandle handle = NULL) : handle_(handle) {}
>
>   ~ScopedCAPIHandle() {
> -    free_(handle_);
> +    reset();
>   }
>
>   void reset(CAPIHandle handle = NULL) {
>     if (handle_ != handle) {
> -      free_(handle_);
> +      FreeProc free_proc;
> +      free_proc(handle_);
>       handle_ = handle;
>     }
>   }
> @@ -91,14 +92,10 @@ class ScopedCAPIHandle {
>
>  private:
>   CAPIHandle handle_;
> -  static const FreeProc free_;
>
>   DISALLOW_COPY_AND_ASSIGN(ScopedCAPIHandle);
>  };
>
> -template<class CH, typename FP>
> -const FP ScopedCAPIHandle<CH, FP>::free_ = FP();
> -
>  template<class CH, typename FP> inline
>  bool operator==(CH h, const ScopedCAPIHandle<CH, FP>& b) {
>   return h == b.get();
>
>
>

[commit-bot: I haz the power, 2011-08-31 04:25:02]

Change committed as 98923

[wtc, 2011-08-31 21:15:28]

LGTM.

http://codereview.chromium.org/7776025/diff/1/crypto/scoped_capi_types.h
File crypto/scoped_capi_types.h (right):

http://codereview.chromium.org/7776025/diff/1/crypto/scoped_capi_types.h#newc...
crypto/scoped_capi_types.h:54: reset();
Just to make sure I understand this correctly: this could
also be written as:
  FreeProc free_proc;
  free_proc(handle_);

Correct?

[Evan Martin, 2011-09-01 21:45:36]

On Wed, Aug 31, 2011 at 2:15 PM,  <wtc@chromium.org> wrote:
> crypto/scoped_capi_types.h:54: reset();
> Just to make sure I understand this correctly: this could
> also be written as:
>  FreeProc free_proc;
>  free_proc(handle_);
>
> Correct?

Yes, I think so.