Bug 149923 – Rationalize JSXXConstructor class definition generated from WebIDL

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore JavaScript

[youenn fablet]

Depending on the WebIDL and options, the JSXXConstructor generated class declaration is changing.
It may be useful to rationalize this so that it can be later templated, which would simplify the binding generator work.

[youenn fablet]

Created attachment 262693 [details]
Patch

[youenn fablet]

(In reply to comment #1)
> Created attachment 262693 [details]
> Patch

mac-wk2 compilation is failing for an unrelated reason, see bug 149751.

[Darin Adler]

Comment on attachment 262693 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=262693&action=review

> Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:4836
> +                my $overloadedIndexString .= $function->{overloadedIndex};

I don’t understand the use of ".=" instead of "=" here.

> Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:5021
> +                push(@$outputArray, "#endif // $conditionalString\n");

I don’t think comments like this one are useful when the entire body of #if/#else/#endif is only 6 lines long. Especially in generated code that people should typically not be reading.

[youenn fablet]

(In reply to comment #3)
> Comment on attachment 262693 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=262693&action=review
> 
> > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:4836
> > +                my $overloadedIndexString .= $function->{overloadedIndex};
> 
> I don’t understand the use of ".=" instead of "=" here.

Right.
Let's also remove $overloadedIndexString at the same time.

> 
> > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:5021
> > +                push(@$outputArray, "#endif // $conditionalString\n");
> 
> I don’t think comments like this one are useful when the entire body of
> #if/#else/#endif is only 6 lines long. Especially in generated code that
> people should typically not be reading.

OK

[youenn fablet]

Created attachment 262764 [details]
Patch for landing

[WebKit Commit Bot]

Comment on attachment 262764 [details]
Patch for landing

Clearing flags on attachment: 262764

Committed r190785: <http://trac.webkit.org/changeset/190785>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Csaba Osztrogonác]

Comment on attachment 262764 [details]
Patch for landing

View in context: https://bugs.webkit.org/attachment.cgi?id=262764&action=review

> Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:5020
> -            push(@$outputArray, "#if $conditionalString\n") if $conditionalString;
>              push(@$outputArray, "ConstructType ${constructorClassName}::getConstructData(JSCell*, ConstructData& constructData)\n");
>              push(@$outputArray, "{\n");
> -            push(@$outputArray, "    constructData.native.function = construct${className};\n");
> -            push(@$outputArray, "    return ConstructTypeHost;\n");
> +            if ($conditionalString) {
> +                push(@$outputArray, "#if $conditionalString\n");
> +                push(@$outputArray, "    constructData.native.function = construct;\n");
> +                push(@$outputArray, "    return ConstructTypeHost;\n");
> +                push(@$outputArray, "#else\n");
> +                push(@$outputArray, "    return Base::getConstructData(cell, constructData);\n");
> +                push(@$outputArray, "#endif\n");

The #else case causes build failure, because there is no cell named variable/argument here.

[Csaba Osztrogonác]

I'm going to fix this in bug149956.