146210 - Need to expose JS_Now-alike via Date.now

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect)

Description

[Mike Shaver (:shaver emeritus)]

Now that we have JS_Now in the API, we should expose it via a Date static as
well, to keep people who are doing timings and the like from having to construct
a new Date object.

Brendan wanted me to do this ages ago, when we first grew JS_Now, but I was
lame.  I'm going to try to cheat this into 1.0, so that it's in the release of
SpiderMonkey that matches it.

Comment 1

[Mike Shaver (:shaver emeritus)]

Attachment: Add Date.now

What could be easier?

Comment 2

[Mike Shaver (:shaver emeritus)]

My ambition knows no bounds.

Comment 3

[Robert Ginda]

Comment on attachment 84637 [details] [diff] [review]
Add Date.now

r=rginda

Comment 4

[Phil Schwartau]

QA NOTE: 

I applied Mike's patch and verified that the new method Date.now()
gives the same result as new Date().valueOf(), that is, the number of 
milliseconds since midnight 01 January, 1970 UTC.


function test()
{
  var now1= new Date().valueOf();
  var now2= Date.now();

  print(now1); 
  print(now2);
}


js> test();
1022094173625
1022094173625

js> test();
1022094174671
1022094174671

js> test();
1022094175656
1022094175656

Comment 5

[Brendan Eich [:brendan]]

Comment on attachment 84637 [details] [diff] [review]
Add Date.now

sr=[e-mail] and I say ship it!

/be

Comment 6

[Mike Shaver (:shaver emeritus)]

chofmann talked sense into me.  1.0 is done like dinner, we'll match 1.5 against
1.0.1?

Comment 7

[Brendan Eich [:brendan]]

Yeah, we should fix the property cache thread safety bug for 1.0.1 too.

/be

Comment 8

[Axel Hecht]

this crashes when called in xpcshell,

js> Date.now()
Assertion failure: lock != NULL, at
/tmp/mozilla/nsprpub/pr/src/pthreads/ptsynch.c:189
Abort

Here's the stack, whatever that should mean:

#5  0xfec357c4 in abort () from /usr/lib/libc.so.1
#6  0xfed96710 in PR_Assert (s=0xfedd7f20 "lock != NULL", 
    file=0xfedd7e50 "/tmp/mozilla/nsprpub/pr/src/pthreads/ptsynch.c", ln=189)
    at /tmp/mozilla/nsprpub/pr/src/io/prlog.c:510
#7  0xfedbee6c in PR_Lock (lock=0x0)
    at /tmp/mozilla/nsprpub/pr/src/pthreads/ptsynch.c:189
#8  0xff2449c4 in JS_dtostr (buffer=0xffbee628 "", bufferSize=25, 
    mode=DTOSTR_STANDARD, precision=0, d=1024655774371)
    at /tmp/mozilla/js/src/jsdtoa.c:2544
#9  0xff28ef40 in js_NumberToString (cx=0x489c8, d=1024655774371)
    at /tmp/mozilla/js/src/jsnum.c:619
#10 0xff2e47f8 in js_ValueToString (cx=0x489c8, v=306506)
    at /tmp/mozilla/js/src/jsstr.c:2577
#11 0xff214ef0 in JS_ValueToString (cx=0x489c8, v=306506)
    at /tmp/mozilla/js/src/jsapi.c:523
#12 0x14f30 in Process (cx=0x489c8, obj=0x49f78, filename=0x0, 
    filehandle=0xfecbc2d4)
    at /tmp/mozilla/js/src/xpconnect/shell/xpcshell.cpp:521
#13 0x15604 in ProcessArgs (cx=0x489c8, obj=0x49f78, argv=0xffbefa00, argc=0)
    at /tmp/mozilla/js/src/xpconnect/shell/xpcshell.cpp:655
#14 0x165cc in main (argc=0, argv=0xffbefa00)
    at /tmp/mozilla/js/src/xpconnect/shell/xpcshell.cpp:910

Date.now() seems to be ok, xpcshell doesn't call InitDtoa and therefor the 
lock isn't created.
How comes that this isn't triggered elsewhere?

Comment 9

[timeless]

The answer is that it is triggered elsewhere :), i knew i'd seen that crash 
before. Oh well, bug 154737.

Comment 10

[Brendan Eich [:brendan]]

Ok, so this patch should go into the 1.0 branch.  I'll mail drivers, since
shaver is AWOL, or AWL, or something.

/be

Comment 11

[chris hofmann]

Comment on attachment 84637 [details] [diff] [review]
Add Date.now

a=chofman for 1.0.1.  add fixed1.0.1 after checking in.

Comment 12

[timeless]

checked into branch (after checkin in bug 154737)

Comment 13

[Phil Schwartau]

Verified on the trunk and the 1.0 branch, the JS shell, and xpcshell.

Date.now() exists and gives the same value as new Date().valueOf();
that is, the number of milliseconds since midnight 01 January, 1970 UTC.