Adding to that, there is also a fair amount of 32-bit assumptions in
the code (such as the assumption that sizeof(int) == sizeof(void*)),
which aren't guaranteed to hold on 64-bit platforms. Fixing these may
turn out to be easy, or it may turn out to require a major
restructuring (I certainly hope not).

Feng, or anyone from the V8 or Chromium team, what's your gut feeling
about this?

- Simon

2008/10/16 Feng Qian <[EMAIL PROTECTED]>:
>
> You missed most complicated part of porting V8 to 64-bit platform:
> garbage collection.
>
> During GC, pointers are encoded using 32-bit address. To make V8 work on 
> 64-bit,
> there might be fair amount of work to get GC right.
>
> We'd like to see V8 run on more platforms, but limited by the time and
> energy, we can
> only support 32-bit and ARM at moment.
>
> On Thu, Oct 9, 2008 at 3:08 PM,  <[EMAIL PROTECTED]> wrote:
>>
>> If V8 devs have good technical reasons why V8 should not run in 64-bit
>> mode (double the pointer size is a pretty flimsy reason, since the
>> vast majority of Javascript programs are too small for it to matter),
>> then that's fine.  However, this is not a good reason for abandoning
>> everyone whose applications are 64-bit for solid technical reasons,
>> eg. thousands of scientific programmers, but who still want to run V8
>> for their scripting.
>>
>> If V8 isn't going to be made available in a 64-bit version, then at
>> the very least a compatibility or 32-bit emulation/interfacing library
>> should be created, so that 64-bit applications can run with 32-bit V8
>> in some way other than by direct linkage.
>>
>> Morgaine.
>> >
>>
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
v8-users mailing list
[email protected]
http://groups.google.com/group/v8-users
-~----------~----~----~----~------~----~------~--~---

Reply via email to