Re: [v8-users] ScriptCompiler::Source ignores ScriptOrigin::ScriptID(). Is this a bug?

2018-06-01 Thread Ben Noordhuis
On Fri, Jun 1, 2018 at 4:22 PM, A.M. wrote: > As for the arbitrary number, it has no usefulness on its own - unless I can > link it back to the script that triggered the exception with this message, > how would you possibly use almost a random value that has no relation to the > script that

Re: [v8-users] ScriptCompiler::Source ignores ScriptOrigin::ScriptID(). Is this a bug?

2018-06-01 Thread A.M.
On Friday, 1 June 2018 04:06:41 UTC-4, Ben Noordhuis wrote: > > You can retrieve the script id after compiling with > v8::UnboundScript::GetId(). (Yes, the casing is inconsistent.) > > Can't use the script object - there are multiple scripts in play because one script "includes" other

Re: [v8-users] why can't SetSecurityToken control cross access?

2018-06-01 Thread Toon Verwaest
As I said: *only* access checked objects can be shared with other contexts. In your case only the object created from |obj_template| is access checked, and hence only *that* object can be shared. In this case you use the template to create the global object underlying the global proxy, so that

Re: [v8-users] why can't SetSecurityToken control cross access?

2018-06-01 Thread fengxingren
But I called SetSecurityToken to set different token for each context. I commented out the SetAccessCheckCallback calls to let access to be controlled only by SetSecurityToken. Even if I add "SetAccessCheckCallback(AccessAlwaysBlocked);" statement, the cross access are still allowed in

Re: [v8-users] why can't SetSecurityToken control cross access?

2018-06-01 Thread Toon Verwaest
If you have two contacts with different security tokens you're only allowed, as an embedder, to leak explicitly access checked objects between the two. Hence only objects that are backed by object templates with access checks. The reason is that we don't check whether you have access for all

Re: [v8-users] ScriptCompiler::Source ignores ScriptOrigin::ScriptID(). Is this a bug?

2018-06-01 Thread Ben Noordhuis
On Thu, May 31, 2018 at 3:31 PM, A.M. wrote: > Thanks, Ben > >> Not a bug, as far as I'm aware. > > How can ths be not a bug - the script source *does take* a script ID and > ignores it? Without this value being passed around, one would have to > compare script names, which is more expensive.

Re: [v8-users] compile hello-world.cc failed

2018-06-01 Thread Ben Noordhuis
On Thu, May 31, 2018 at 2:30 PM, czczcheng wrote: > args.gn(get the static library): > is_debug = false > target_cpu = "x64" > is_component_build = false > v8_static_library = true > use_custom_libcxx = false > use_custom_libcxx_for_host = false > use_sysroot=false > is_clang=false > > > using

Re: [v8-users] why can't SetSecurityToken control cross access?

2018-06-01 Thread Ben Noordhuis
On Thu, May 31, 2018 at 3:52 AM, wrote: > ur... I still can not get it. In the following modification, I changed > object x by scripts. > > Modification Three: > > TEST(EvalInAccessCheckedContext) { > v8::Isolate* isolate = CcTest::isolate(); > v8::HandleScope scope(isolate); > >