Bug 4258 – Date().toString() only includes GMT offset, not timezone string

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[John Payne]

So far this has only affected a company internal website that parse the datestring to get the timezone.

Internet Explorer, Netscape and Mozilla all have different ways of reporting the timezone.  I suggest 
adopting the Mozilla format, as that is closest to the rest of the output of Date().toString()

Safari currently reports:
Mon Aug 01 2005 15:21:19 GMT-0400

Mozilla (firefox) reports:
Mon Aug 01 2005 15:21:22 GMT-0400 (EDT)

It feels like you could use a lookup into KnownZone in formatTime() to accomplish this.

[John Payne]

just updating OS to 10.4 and version to 412+ to highlight that this is also an issue on tiger... just in case 
things are being prioritised away from jaguar bugs :)

[Justin Haygood]

Created attachment 5404 [details]
Makes JavaScriptCore match Mozilla

Makes JavaScriptCore match Mozilla if the timezone is known.

[Justin Haygood]

Comment on attachment 5404 [details]
Makes JavaScriptCore match Mozilla

Moves KnownZones higher so that formatTime can find it, changes formatTime to
make use of it.

[John Payne]

from a checkout today...

/Users/jpayne/src/webkit/JavaScriptCore/kjs/date_object.cpp: In function 'KJS::UString KJS::formatTime
(const tm&)':
/Users/jpayne/src/webkit/JavaScriptCore/kjs/date_object.cpp:245: warning: suggest parentheses 
around assignment used as truth value
/Users/jpayne/src/webkit/JavaScriptCore/kjs/date_object.cpp:247: error: 'gmtOffset' was not declared 
in this scope
/Users/jpayne/src/webkit/JavaScriptCore/kjs/date_object.cpp:263: error: name lookup of 'i' changed 
for new ISO 'for' scoping
/Users/jpayne/src/webkit/JavaScriptCore/kjs/date_object.cpp:245: error:   using obsolete binding at 'i'
/Users/jpayne/src/webkit/JavaScriptCore/kjs/date_object.cpp:263: warning: format '%s' expects type 
'char*', but argument 10 has type 'const KJS::KnownZone'
** BUILD FAILED **                                                              

[Darin Adler]

Comment on attachment 5404 [details]
Makes JavaScriptCore match Mozilla

It seems strange to list the time zone only if it's one of the hardcoded "known
time zones".

On Mac OS X and many other Unix systems, the timezone() function can be used to
get the time zone abbreviation. On Win32, the GetTimeZoneInformation() function
can probably be used.

[Maciej Stachowiak]

I agree with darin's suggestion. Please use a function that knows about more time zones. Also, please 
provide a test case for this, if it is not already covered by the JS tests. It may be tricky to write a layout test 
for this that is time zone independent.

[Rob Buis]

Created attachment 9500 [details]
First attempt

When in utc mode, I just use tm_zone info here. This should be ok since localtime
should set it up correctly. Unfortunately there seem to be no tests for this. I used
http://sackheads.org/~jpayne/timestring.html to test manually, and it matches Mozilla.
Let me know if any tests should be added or whether we can reuse existing tests.
Cheers,

Rob.

[Darin Adler]

Comment on attachment 9500 [details]
First attempt

Looks fine to me, but doesn't this affect some of our tests?

r=me if not

[Rob Buis]

Hi Darin,

(In reply to comment #8)
> (From update of attachment 9500 [details] [edit])
> Looks fine to me, but doesn't this affect some of our tests?
> 
> r=me if not

bdash couldn't see any changed results. But we concluded, it may be best
if someone in the PDC timezone checks it, as NZ timezone gives issues with some
of the tests.
Cheers,

Rob.

[Darin Adler]

This change broke the Windows buildbot.