Cog Closure Compiler

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Cog Closure Compiler

Lukas Renggli
Hi,

I finally found out where to get the image level code for the Cog Closure Image.

     http://www.mirandabanda.org/files/Cog/Closures0811/Bootstrap/

I used the latest Squeak 4.0 VM from John and I tried to load the
change-sets into the latest Pharo, as well as into the latest Squeak
3.9.1. In both cases it breaks pretty early on.

It is easy to add empty dummy classes that the script tries to patch
(e.g. DecompilerTest) but then there are dozens of undeclared
variables reported on transcript and the file-in eventually dies. The
fact that debugger, compiler, browser and compiled methods are then in
a strange state doesn't make it easier ...

Has anybody had more success with this? Anybody willing to help making
this thing work in Pharo?

I desire fast closures so badly ... ;-)

Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: Cog Closure Compiler

keith1y
Lukas Renggli wrote:

> Hi,
>
> I finally found out where to get the image level code for the Cog Closure Image.
>
>      http://www.mirandabanda.org/files/Cog/Closures0811/Bootstrap/
>
> I used the latest Squeak 4.0 VM from John and I tried to load the
> change-sets into the latest Pharo, as well as into the latest Squeak
> 3.9.1. In both cases it breaks pretty early on.
>
> It is easy to add empty dummy classes that the script tries to patch
> (e.g. DecompilerTest) but then there are dozens of undeclared
> variables reported on transcript and the file-in eventually dies. The
> fact that debugger, compiler, browser and compiled methods are then in
> a strange state doesn't make it easier ...
>
> Has anybody had more success with this? Anybody willing to help making
> this thing work in Pharo?
>
> I desire fast closures so badly ... ;-)
>
> Lukas
>  
I cant help thinking that atomic loading would be useful here. I assume
that Elliot has not used traits, so it "should work", however:

There is an image here with lpf and atomic loading enabled:

http://bob.warwick.st/3.11/Squeak3.10.2-lpf-atomic/090305-0113/

When I tried to add stuff to this image I get a load of obsolete classes
that are not tidied up, as can be seen in this image here:

http://bob.warwick.st/output/BuildBase/Squeak3.10.2-build-atomic/090305-0135/

So honestly it might be worth someone's time to have a look at it, and
find out what matthew broke :-)

Keith






_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project