631132 - Expose public APIs for converting doubles to signed/unsigned int per ECMA spec

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Boris Zbarsky [:bzbarsky]]

Could be useful in places where we have to write bindings.

Comment 1

[Andreas Gal :gal]

Attachment: patch

Comment 2

[Andreas Gal :gal]

NPOTB, not currently in use.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 509339 [details] [diff] [review]
patch

Land it.

Comment 4

[Brendan Eich [:brendan]]

Comment on attachment 509339 [details] [diff] [review]
patch

Does it not seem confusing to have the old JS_ValueToInt32, which rounds to nearest and fails if the double is out of range, the newer but still ancient JS_ValueToECMA{I,Ui}nt32 pair, and now these JS_DoubleTo{I,Ui}nt32 which match ECMA-262 and do not match JS_ValueToInt32?

We could rotate names to give the old, pre-ECMA JS_ValueToInt32 a long and better name, giving the ECMA names the short and "ECMA"-free names. We could mangle the names of the JS_DoubleTo{I,Ui}nt32 APIs to contain "ECMA". But just adding more mixed naming does not seem good. Right?

/be

Comment 5

[Jeff Walden [:Waldo]]

Hyperbole for sure, but rotating names feels somewhat like rearranging deck chairs to me.  The names certainly aren't pretty, but name stability seems to trump aesthetics to me, as far as external APIs go.  (Although fixing bugs/enabling better features itself trumps stability, cf. operation callback breakages and so on.)  If we want to regularize names, it seems most sensible to do that as part of JSAPI2.

Comment 6

[Brendan Eich [:brendan]]

Agree with comment 5, which leaves adding "ECMA" to the parallel DoubleTo names. Agreed?

/be

Comment 7

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

This is blocking some of my work. Any chance this could be landed soonish?

Comment 8

[Andreas Gal :gal]

http://hg.mozilla.org/tracemonkey/rev/8359361f8a7b

Comment 9

[Boris Zbarsky [:bzbarsky]]

What about comment 4 through comment 6?

Comment 10

[Jeff Walden [:Waldo]]

Pre-landing I'm willing to care about names (or not care, as my previous comment seems to say).  Post-landing, whatever got landed is good enough, since naming consistency is basically impossible until JSAPI2.

Comment 11

[Andreas Gal :gal]

Ah ****. I overlooked the post review comments. I am happy with any naming. Waldo, guide me. What should I name how? I will do a little follow-up before I merge to m-c.

Comment 12

[Chris Leary [:cdleary] (not checking bugmail)]

http://hg.mozilla.org/mozilla-central/rev/8359361f8a7b

Comment 13

[Chris Leary [:cdleary] (not checking bugmail)]

Whoops, sorry, this wasn't merged. Reopened.

Comment 14

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Yes it did?

http://hg.mozilla.org/mozilla-central/file/1a89509e25e4/js/src/jsapi.cpp#l523

Comment 15

[Eric Shepherd [:sheppy]]

Documented:

https://developer.mozilla.org/en/SpiderMonkey/JSAPI_Reference/JS_DoubleToInt32

Also mentioned on Firefox 5 for developers.