Bug 39972 – Fix style errors of dtoa

• Component: Web Template Framework
• Status: RESOLVED FIXED

[Kent Tamura]

Fix style errors of dtoa

[Kent Tamura]

Created attachment 57498 [details]
Patch

[Shinichiro Hamaji]

Comment on attachment 57498 [details]
Patch

JavaScriptCore/wtf/dtoa.cpp:210
 +  static ALWAYS_INLINE uint32_t* storeInc(uint32_t* p, uint16_t high, uint16_t low)
I think this is a refactoring rather than a style fix. I guess we can remove the style error just by adding spaces after commas?

[Kent Tamura]

Thank you for looking at the patch.

(In reply to comment #2)
> (From update of attachment 57498 [details])
> JavaScriptCore/wtf/dtoa.cpp:210
>  +  static ALWAYS_INLINE uint32_t* storeInc(uint32_t* p, uint16_t high, uint16_t low)
> I think this is a refactoring rather than a style fix. I guess we can remove the style error just by adding spaces after commas?

It's also a style fix though check-webkit-style doesn't complain it.

http://webkit.org/coding/coding-style.html
> 11. Prefer const to #define. Prefer inline functions to macros.

[Shinichiro Hamaji]

> It's also a style fix though check-webkit-style doesn't complain it.
> 
> http://webkit.org/coding/coding-style.html
> > 11. Prefer const to #define. Prefer inline functions to macros.

I see. So, we should fix rounded_product and rounded_quotient as well?

I guess we can use reference if there are no performance regression (I'm guessing the compiler is clever enough)?

static ALWAYS_INLINE void storeInc(uint32_t*& p, uint16_t high, uint16_t low)
{
  // ...
  p++;
}

[Kent Tamura]

(In reply to comment #4)
> I see. So, we should fix rounded_product and rounded_quotient as well?

Yes, we should.  But I didn't change them because I thought they could be member functions of type U.

> I guess we can use reference if there are no performance regression (I'm guessing the compiler is clever enough)?
> 
> static ALWAYS_INLINE void storeInc(uint32_t*& p, uint16_t high, uint16_t low)
> {
>   // ...
>   p++;
> }

I don't like to change to the reference unless you have a strong objection.  I believe returning updated pointer has better code readability.

[Shinichiro Hamaji]

Comment on attachment 57498 [details]
Patch

> I don't like to change to the reference unless you have a strong objection.  I believe returning updated pointer has better code readability.

Hmm I slightly prefer the reference so that we won't forget to assign the return value, but it's not a strong objection. Please feel free to land this patch as is.

[WebKit Commit Bot]

Comment on attachment 57498 [details]
Patch

Clearing flags on attachment: 57498

Committed r60468: <http://trac.webkit.org/changeset/60468>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.