610228 - Fix warnings when embedding

Status: RESOLVED WORKSFORME

(Core :: JavaScript Engine, defect)

Description

[Philip Taylor]

Attachment: fix some warnings

When embedding SpiderMonkey (from tracemonkey 636b1e5a994f) in an otherwise warning-clean application which uses flags including -Wunused-parameter and -Wredundant-decls, I get the following warnings:

  jsapi.h:3323: warning: unused parameter ‘cx’
  jsapi.h:3364: warning: unused parameter ‘cx’
  jsobj.h:1293: warning: unused parameter ‘cx’
  jsvalue.h:526: warning: unused parameter ‘why’
  jsvalue.h:946: warning: redundant redeclaration of ‘JSBool (* js::Valueify(JSBool (*)(JSContext*, JSObject*, const jsval*, JSBool*)))(JSContext*, JSObject*, const js::Value*, JSBool*)’ in same scope
  jsvalue.h:942: warning: previous declaration of ‘JSBool (* js::Valueify(JSBool (*)(JSContext*, JSObject*, const jsval*, JSBool*)))(JSContext*, JSObject*, const js::Value*, JSBool*)’
  jsvalue.h:947: warning: redundant redeclaration of ‘JSBool (* js::Jsvalify(JSBool (*)(JSContext*, JSObject*, const js::Value*, JSBool*)))(JSContext*, JSObject*, const jsval*, JSBool*)’ in same scope
  jsvalue.h:943: warning: previous declaration of ‘JSBool (* js::Jsvalify(JSBool (*)(JSContext*, JSObject*, const js::Value*, JSBool*)))(JSContext*, JSObject*, const jsval*, JSBool*)’

Comment 1

[David Mandelin [:dmandelin]]

Do you need a reviewer for this?

Comment 2

[Philip Taylor]

Oops, yes. (I always forget to do this properly.)

Comment 3

[Philip Taylor]

Attachment: fix some more warnings

Updated to fix warnings on Windows too.

Comment 4

[Jeff Walden [:Waldo]]

Comment on attachment 489277 [details] [diff] [review]
fix some more warnings

Shaver's not reviewing patches so much these days, I think.

As far as the patch goes:

Should there be a JS_MARK_UNUSED macro instead of the (void)cx stuff?  That seems a little nicer, cuts out the oft-repeated comment.

The win32-ifdefs should be as narrowly scoped as possible.  I'm pretty sure this can be scoped a little narrower -- to just JSValueType and friends, roughly.

As far as the jsvalue.h pragmas go, those we should just convert to == JS_TRUE (assuming that's what the problem is).

Then again, a lot of code's changed since this patch was posted.  How much of it still applies, and is needed?  Whip up an adjusted patch and I'll review quickly.

Comment 5

[Atul Aggarwal]

No warning on current code on linux using gcc 4.6.1 in jsapi.h, jsobj.h and jsvalue.h.
Suggest closing this bug.