352856 - Provide guaranteed way to create a JSString from a UTF8 char*

Status: RESOLVED DUPLICATE of bug 315288

(Core :: JavaScript Engine, defect)

Description

[Boris Zbarsky [:bzbarsky]]

Right now we have several cases (xpcconvert, nsJSNPRuntime) where we do a UTF8->UTF16 conversion in Gecko, then call JS_NewUCString on the result.  With the recent UTF8 work on the JS engine, is it possible to use JS_NewStringCopy* or something instead?  Is there a version that guarantees that it is and will remain UTF8?

Comment 1

[Brendan Eich [:brendan]]

See bug 232182, which morphed and was closed, but still depends on the real bug to fix here: bug 315288.

Asking question in a bug summary is not good use of the bugsystem IMHO.  I think this is a dup of bug 315288.  But I'll make this bug depend on that bug instead, in case I am missing something and being grumpy for no good reason :-/.

/be

Comment 2

[Brendan Eich [:brendan]]

Note that even with JS_C_STRINGS_ARE_UTF8 defined, any use of the char (byte sized character) JS string API (e.g., JS_NewString) will still decode into UTF-16 for storage in a jschar (uint16) buffer.

If this bug is really asking JS to store UTF-8, not just support it via the char string API, the summary should be fixed.

/be

Comment 3

[Boris Zbarsky [:bzbarsky]]

> I think this is a dup of bug 315288

If so, we should file separate bugs on the Gecko code that could use this...

Let me resummarize to make it clearer what I want out of the JS engine here.  ;)  I just want to pass in a UTF8 char* and then have the JS engine deal with conversion to UTF16 (actual conversion, not byte-inflation).  Then I can remove various Gecko code that does that manually right now.  But I don't want to do that until there's an absolute guarantee that whatever JS engine API I'm using is UTF8 and will stay that way.

Comment 4

[Brendan Eich [:brendan]]

(In reply to comment #3)
> > I think this is a dup of bug 315288
> 
> If so, we should file separate bugs on the Gecko code that could use this...
> 
> Let me resummarize to make it clearer what I want out of the JS engine here. 
> ;)  I just want to pass in a UTF8 char* and then have the JS engine deal with
> conversion to UTF16 (actual conversion, not byte-inflation).

Then this is a dup of bug 315288.

> Then I can remove
> various Gecko code that does that manually right now.  But I don't want to do
> that until there's an absolute guarantee that whatever JS engine API I'm using
> is UTF8 and will stay that way.

Filing this bug does not make the JS_C_STRINGS_ARE_UTF8 any more absolute than it already is.  Its name says what it does unambiguously given brevity needs.

/be

*** This bug has been marked as a duplicate of 315288 ***