Bug 171240 – [Win] Use Clang's __has_declspec_attribute for export macros

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Don Olmstead]

Currently a OS(WINDOWS) check is used to determine whether __declspec should be used for exporting symbols in shared libraries. Clang defines a macro __has_declspec_attribute that can be used to determine if the platform supports __declspec.

[Don Olmstead]

Created attachment 307995 [details]
Patch

[Build Bot]

Attachment 307995 [details] did not pass style-queue:


ERROR: Source/WTF/wtf/Compiler.h:54:  One space before end of line comments  [whitespace/comments] [5]
ERROR: Source/WTF/wtf/Compiler.h:54:  Should have a space between // and comment  [whitespace/comments] [4]
Total errors found: 2 in 6 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Alex Christensen]

Comment on attachment 307995 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=307995&action=review

> Source/WTF/wtf/Platform.h:440
> +#if OS(WINDOWS) \
> +    || (COMPILER_HAS_CLANG_DECLSPEC(dllimport) && COMPILER_HAS_CLANG_DECLSPEC(dllexport))

These can be on one line.
Also, if it has __declspec(dllimport) can't we assume it has __declspec(dllexport)?

> Source/WTF/wtf/Platform.h:441
> +#define USE_DECLSPEC_ATTRIBUTE 1

__declspec is a keyword according to https://msdn.microsoft.com/en-us/library/dabb5z75.aspx
I wish clang had used __has_declspec_keyword, but they didn't.

> Source/WTF/wtf/Platform.h:443
> +#elif defined(__GNUC__) && !defined(__CC_ARM) && !defined(__ARMCC__)

Is there no visibility attribute on ARM?

[Don Olmstead]

(In reply to Alex Christensen from comment #3)
> Comment on attachment 307995 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=307995&action=review
> 
> > Source/WTF/wtf/Platform.h:440
> > +#if OS(WINDOWS) \
> > +    || (COMPILER_HAS_CLANG_DECLSPEC(dllimport) && COMPILER_HAS_CLANG_DECLSPEC(dllexport))
> 
> These can be on one line.
> Also, if it has __declspec(dllimport) can't we assume it has
> __declspec(dllexport)?

I would assume that would be the case but just made it completely explicit. Can change that.

> 
> > Source/WTF/wtf/Platform.h:441
> > +#define USE_DECLSPEC_ATTRIBUTE 1
> 
> __declspec is a keyword according to
> https://msdn.microsoft.com/en-us/library/dabb5z75.aspx
> I wish clang had used __has_declspec_keyword, but they didn't.
> 
> > Source/WTF/wtf/Platform.h:443
> > +#elif defined(__GNUC__) && !defined(__CC_ARM) && !defined(__ARMCC__)
> 
> Is there no visibility attribute on ARM?

That line has actually been there since the first commit of ExportMacros.h so its over 6 years old. I'm not sure the initial thinking of it. Should we just remove it?

[Alex Christensen]

Comment on attachment 307995 [details]
Patch

Let's just do this as it is now.

[WebKit Commit Bot]

Comment on attachment 307995 [details]
Patch

Clearing flags on attachment: 307995

Committed r215766: <http://trac.webkit.org/changeset/215766>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.