I am working on a high-quality static compiler that currently accepts Java class files as input. I have a Masters student looking for a project, and I was thinking that it might not be *too* much work to re-target VMMaker to generate Java or Java class files. This would possibly make squeak even more portable, and would provide a good test-bed for my compiler.
Opinions (pro and con, technical and philosophical) welcome.
I think I also noticed that someone was essentially running Smalltalk code directly on a Java VM. I'd be interested in pointers to that work too.
I've set the reply-to to me, because I don't think this is likely of interest to many other people at present, but if you're interested let me know and I'll either post a followup or send you an email when I've gathered replies.
Thanks ../Dave
Hello Dave,
DM> I've set the reply-to to me, because I don't think this is likely of DM> interest to many other people at present, but if you're interested let
personally I'd prefer the discussion take place here on Squeak dev.
Cheers
Herbert mailto:herbertkoenig@gmx.net
In message 20050917113345.E874F83836@sarg.ryerson.ca Dave Mason dmason@sarg.ryerson.ca wrote:
I am working on a high-quality static compiler that currently accepts Java class files as input. I have a Masters student looking for a project, and I was thinking that it might not be *too* much work to re-target VMMaker to generate Java or Java class files. This would possibly make squeak even more portable, and would provide a good test-bed for my compiler.
Opinions (pro and con, technical and philosophical) welcome.
Urk. Please, no.
For a start, it wouldn't make Squeak even more portable (to the best of my knowledge) since Squeak has been ported to just about every machine that ever existed. I'm not entirely certain but I think the Difference Engine at the London Science Museum managed 3+4 doit :-)
Next, consider the appalling overheads. So far as I can work out the typical java vm and support libraries appear to occupy many hundreds of megabytes.
<sounds of tim banging head on desk and weeping 'make it go away, oh the pain, the humanity'>
tim -- Tim Rowledge, tim@rowledge.org, http://www.rowledge.org/tim In computer science, we stand on each other's feet. - Brian Reid
On Sat, Sep 17, 2005 at 08:41:38AM -0700, Tim Rowledge wrote:
In message 20050917113345.E874F83836@sarg.ryerson.ca Dave Mason dmason@sarg.ryerson.ca wrote:
I am working on a high-quality static compiler that currently accepts Java class files as input. I have a Masters student looking for a project, and I was thinking that it might not be *too* much work to re-target VMMaker to generate Java or Java class files. This would possibly make squeak even more portable, and would provide a good test-bed for my compiler.
Opinions (pro and con, technical and philosophical) welcome.
Urk. Please, no.
For a start, it wouldn't make Squeak even more portable (to the best of my knowledge) since Squeak has been ported to just about every machine that ever existed. I'm not entirely certain but I think the Difference Engine at the London Science Museum managed 3+4 doit :-)
Next, consider the appalling overheads. So far as I can work out the typical java vm and support libraries appear to occupy many hundreds of megabytes.
<sounds of tim banging head on desk and weeping 'make it go away, oh the pain, the humanity'>
but it *would* be nice to do a BCPL code generator for slang, don't you think?
Dave
"Dave" == Dave Mason dmason@sarg.ryerson.ca writes:
Dave> I am working on a high-quality static compiler that currently Dave> accepts Java class files as input. I have a Masters student Dave> looking for a project, and I was thinking that it might not be Dave> *too* much work to re-target VMMaker to generate Java or Java Dave> class files. This would possibly make squeak even more Dave> portable, and would provide a good test-bed for my compiler.
"not *too* much work" seems wrong to me. Sure, everything can be done. But when you reach pointers manipulation, you will likely end up with ugly byte arrays manipulation in Java unless you redesign whole parts of the system.
Dave> Opinions (pro and con, technical and philosophical) welcome.
Doing it might be a lot of fun, especially as a proof of concept, and could lead to an interesting conference article. However, trying to base the future of Squeak on a JVM would be totally wrong IMO as Squeak is probably already more portable than Java (are there many places where Squeak doesn't run and where a full JVM does?). Also, the part of the JVM you would need isn't likely to fit in small embedded systems while Squeak does.
Sam
squeak-dev@lists.squeakfoundation.org