Closed Bug 290114 Opened 20 years ago Closed 20 years ago

Improve GC_MARK_DEBUG handling of NaN/INF to avoid PAR: Floating-point NaN

Categories

(Core :: JavaScript Engine, enhancement)

x86
Windows XP
enhancement
Not set
normal

Tracking

()

VERIFIED WONTFIX

People

(Reporter: timeless, Assigned: timeless)

Details

steps:
1. build w/ gc_mark_debug
2. run xpcshell under purify
3. quit

actual results:
[W] PAR: Floating-point NaN at (0x13e660) {3 occurrences}
        Call location
            fprintf        [f:\vs70builds\3077\vc\crtbld\crt\src\fprintf.c:48]
            gc_dump_thing  [r:\cenzic\mozilla\js\src\jsgc.c:935]
                      {
                        JSXMLQName *qn = (JSXMLQName *)thing;
                        fprintf(fp, "qname %s(%s):%s",
                                JS_GetStringBytes(qn->prefix),
JS_GetStringBytes(qn->uri),
                                JS_GetStringBytes(qn->localName));
                        break;
                      }
                      case GCX_XML:
                      {
                        extern const char *js_xml_class_str[];
                        JSXML *xml = (JSXML *)thing;
                        fprintf(fp, "xml %8p %s", xml,
js_xml_class_str[xml->xml_class]);
                        break;
                      }
                      case GCX_DOUBLE:
             =>         fprintf(fp, "double %g", *(jsdouble *)thing);

prefered results:
048fcb80 NaN -1 via
atom(AnyName).__proto__(Object).__proto__(Object).__parent__(JSDGlobal).NaN(double).
048fcb88 +INF 0 via
atom(AnyName).__proto__(Object).__proto__(Object).__parent__(JSDGlobal).Infinity(double).
048fd9c0 -INF 0 via
atom(AnyName).__proto__(Object).__proto__(Object).__parent__(JSDGlobal).Number(Function).NEGATIVE_INFINITY(double).
I really don't want more hacks in the code to avoid these bogus impurities.  Why
don't you configure purify to ignore 'em?

/be
Status: UNCONFIRMED → RESOLVED
Closed: 20 years ago
Resolution: --- → WONTFIX
Summary: Improve GC_MARK_DEBUG handling of NaN/INF to avoid PAR: Floating-point NaN → Improve GC_MARK_DEBUG handling of NaN/INF to avoid PAR: Floating-point NaN
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.