Enable isolate tests with leg and legium.

Enable isolate tests with leg and legium.
Committed: https://code.google.com/p/dart/source/detail?r=5411

[ngeoffray, 2012-03-12 23:18:36]

https://chromiumcodereview.appspot.com/9677019/diff/1/frog/leg/lib/uri_toremo...
File frog/leg/lib/uri_toremove.dart (right):

https://chromiumcodereview.appspot.com/9677019/diff/1/frog/leg/lib/uri_toremo...
frog/leg/lib/uri_toremove.dart:1: // Copyright (c) 2012, the Dart project
authors.  Please see the AUTHORS file
This file should soon be removed. The original uri file is blocking all isolate
tests, so duplicating here allows us to make progress.

What is blocking is the const constructor call:
const RegExp('.....')

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
File lib/isolate/frog/isolateimpl.dart (right):

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:203: // Currently, none of the two Dart
classes have subclasses.
This is more of a cleanup than a hack IMO. Siggi, Eli let me know what you
think.

[Siggi Cherem (dart-lang), 2012-03-12 23:40:57]

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
File lib/isolate/frog/isolateimpl.dart (right):

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:203: // Currently, none of the two Dart
classes have subclasses.
On 2012/03/12 23:18:36, ngeoffray wrote:
> This is more of a cleanup than a hack IMO. Siggi, Eli let me know what you
> think.

looks good to me. I'm working on adding Timer to corelib, which is basically
based on this. Hopefully then we can hide this code behind the Timer interface. 

I don't plan to implement Timer outside the browser, so we will still need to
detect if it is available, for instance checking if new Timer(x, callback)
throws an exception.

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:206: int setTimeout(_TimeoutHandler handler,
int timeout) native;
is this _TimeoutHandler needed, or can we simply do:
int setTimeout(void handler(), int timeout) native;
?

[ngeoffray, 2012-03-13 12:51:37]

Thanks Siggi

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
File lib/isolate/frog/isolateimpl.dart (right):

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:203: // Currently, none of the two Dart
classes have subclasses.
On 2012/03/12 23:40:58, sigmund wrote:
> On 2012/03/12 23:18:36, ngeoffray wrote:
> > This is more of a cleanup than a hack IMO. Siggi, Eli let me know what you
> > think.
> 
> looks good to me. I'm working on adding Timer to corelib, which is basically
> based on this. Hopefully then we can hide this code behind the Timer
interface. 

Sounds good.

> 
> I don't plan to implement Timer outside the browser, so we will still need to
> detect if it is available, for instance checking if new Timer(x, callback)
> throws an exception.

I don't understand: if you're adding something to the corelib, it ought to be
available to both browser and standalone. Or are you saying that the actual
implementation of Timer in standalone will be implemented by someone else?

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:206: int setTimeout(_TimeoutHandler handler,
int timeout) native;
On 2012/03/12 23:40:58, sigmund wrote:
> is this _TimeoutHandler needed, or can we simply do:
> int setTimeout(void handler(), int timeout) native;
> ?

Leg does not handle function type parameters very well right now. We're fixing
it.

[kasperl, 2012-03-13 13:44:42]

LGTM.

[Siggi Cherem (dart-lang), 2012-03-14 23:23:12]

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
File lib/isolate/frog/isolateimpl.dart (right):

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:203: // Currently, none of the two Dart
classes have subclasses.
On 2012/03/13 12:51:37, ngeoffray wrote:
> On 2012/03/12 23:40:58, sigmund wrote:
> > On 2012/03/12 23:18:36, ngeoffray wrote:
> > > This is more of a cleanup than a hack IMO. Siggi, Eli let me know what you
> > > think.
> > 
> > looks good to me. I'm working on adding Timer to corelib, which is basically
> > based on this. Hopefully then we can hide this code behind the Timer
> interface. 
> 
> Sounds good.
> 
> > 
> > I don't plan to implement Timer outside the browser, so we will still need
to
> > detect if it is available, for instance checking if new Timer(x, callback)
> > throws an exception.
> 
> I don't understand: if you're adding something to the corelib, it ought to be
> available to both browser and standalone. Or are you saying that the actual
> implementation of Timer in standalone will be implemented by someone else?

It would be supported standalone in the VM, but not when compiling with frog to
run in d8. In nodejs we had setTimeout available, but I haven't found something
equivalent for d8. I was leaning to say that in such case Timer is not
available.

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:206: int setTimeout(_TimeoutHandler handler,
int timeout) native;
On 2012/03/13 12:51:37, ngeoffray wrote:
> On 2012/03/12 23:40:58, sigmund wrote:
> > is this _TimeoutHandler needed, or can we simply do:
> > int setTimeout(void handler(), int timeout) native;
> > ?
> 
> Leg does not handle function type parameters very well right now. We're fixing
> it.

good to know. Thx.

[ngeoffray, 2012-03-15 09:30:56]

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
File lib/isolate/frog/isolateimpl.dart (right):

https://chromiumcodereview.appspot.com/9677019/diff/1/lib/isolate/frog/isolat...
lib/isolate/frog/isolateimpl.dart:203: // Currently, none of the two Dart
classes have subclasses.
On 2012/03/14 23:23:12, sigmund wrote:
> On 2012/03/13 12:51:37, ngeoffray wrote:
> > On 2012/03/12 23:40:58, sigmund wrote:
> > > On 2012/03/12 23:18:36, ngeoffray wrote:
> > > > This is more of a cleanup than a hack IMO. Siggi, Eli let me know what
you
> > > > think.
> > > 
> > > looks good to me. I'm working on adding Timer to corelib, which is
basically
> > > based on this. Hopefully then we can hide this code behind the Timer
> > interface. 
> > 
> > Sounds good.
> > 
> > > 
> > > I don't plan to implement Timer outside the browser, so we will still need
> to
> > > detect if it is available, for instance checking if new Timer(x, callback)
> > > throws an exception.
> > 
> > I don't understand: if you're adding something to the corelib, it ought to
be
> > available to both browser and standalone. Or are you saying that the actual
> > implementation of Timer in standalone will be implemented by someone else?
> 
> It would be supported standalone in the VM, but not when compiling with frog
to
> run in d8. In nodejs we had setTimeout available, but I haven't found
something
> equivalent for d8. I was leaning to say that in such case Timer is not
> available.

Oh I see. So I guess you want to throw a runtime exception by doing something
like:

class Timer {
   Timer() {
      if (JS_CODE("typeof window == 'undefined')) {
        throw 'Timer not supported';
      }
      ....
   }
}