Bug 138934 – Add some comments to describe the DFG UseKind representations.

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Mark Lam]

Also regrouped the UseKind enums by representation.

[Mark Lam]

Created attachment 241980 [details]
the patch.

[Geoffrey Garen]

Comment on attachment 241980 [details]
the patch.

View in context: https://bugs.webkit.org/attachment.cgi?id=241980&action=review

> Source/JavaScriptCore/dfg/DFGUseKind.h:38
>      UntypedUse,

Which case is UntypedUse? (Seems like it should be JSValue, right?)

> Source/JavaScriptCore/dfg/DFGUseKind.h:43
> +    //    in a FP register.

"an"

[Mark Lam]

Created attachment 241989 [details]
patch 2: addressed Geoff's feedback.

[Filip Pizlo]

Comment on attachment 241989 [details]
patch 2: addressed Geoff's feedback.

View in context: https://bugs.webkit.org/attachment.cgi?id=241989&action=review

> Source/JavaScriptCore/dfg/DFGUseKind.h:54
>      UntypedUse,

This could be dangerous; I believe that UntypedUse is expected to be zero.

[Filip Pizlo]

Comment on attachment 241989 [details]
patch 2: addressed Geoff's feedback.

Please arrange to have UntypedUse be the first thing, as before.

[Mark Lam]

Created attachment 241992 [details]
patch 3: UntypedUse is at enum 0 again, with extra comments.

[Mark Lam]

Created attachment 241995 [details]
patch 4: don't like the comment about UntypedUse in last patch.  Updated.

[Filip Pizlo]

Comment on attachment 241995 [details]
patch 4: don't like the comment about UntypedUse in last patch.  Updated.

You could also have moved all of the other JS representations up with UntypedUse.  That would make the most logical sense to me.

[Mark Lam]

Created attachment 241996 [details]
patch 5: moved al the JSValue representation UseKinds up nelow UntypedUse as Filip suggested.

[Mark Lam]

Thanks for the review.  Landed in r176425: <http://trac.webkit.org/r176425>.