4 ms·

Christ, this guy never ceases to amaze. His talk at I/O a couple years ago was absolutely fantastic: http://www.youtube.com/watch?v=BttI-y9VzXQ http://www.you

Christ, this guy never ceases to amaze. His talk at I/O a couple years ago was absolutely fantastic: http://www.youtube.com/watch?v=BttI-y9VzXQ http://www.youtube.com/watch?v=BttI-y9VzXQ

moron4hire 16y ago
One of the points he makes is that VMs are obvious for the purpose of language interop. Seems like C and Lisp (and D, another language he mentions) have been interoperating with each other without a VM in common for quite a while.
- Goladus 16y ago
  But it's not a many to many mapping, that's the point he was making. Lots of languages interop with C, but you can't just import a ruby module into a python script very easily. In fact if you want to link to C libraries from D you have to modify the C header files to turn them into D modules.
  - moron4hire 16y ago
    Sure, but it doesn't seem intractable to be able to design native languages to be able to interop. Lisp has been around since the late 50s, C since the 70s. The fact that they interop at all is kinda awesome.
  - andrewf 16y ago
    I think Microsoft actually did a reasonable job with the "original" COM (ie OLE 2.0 - the stuff that let you put an Excel 95 spreadsheet into a Word 95 document). It had a bit too much bolted on, probably due to its origins in OLE, and I think "dynamic OLE" was a tragic mistake. And its reference counting approach was a lot more appropriate in the early 90s than it is today (where a compacting GC may be the smart choice, performance-wise). But for making calls between objects in C, C++, VB6, Delphi, and pretty much every other language that lived on the Windows platform, it worked really well.
    - limmeau 16y ago
      For a compacting GC, you'd have to hunt down a complete set of references from running programs written in unknown languages, with strange errors if the set is underapproximated even once. I'd rather have reference counting.
- nickik 16y ago
  It works but it is WAY HARD. Interact from ruby to common lisp. On a VM its easy a call from Jruby to clojure is not to hard. Its stupid to implement your own GC with every knew language and there are way more things that every C-Based language has to reimplement. Thats why it always took around 10 years until a language got really good and fast. Today you can have a language in 2 years with the biggest set of libs out there.

Christ, this guy never ceases to amaze. His talk at I/O a couple years ago was absolutely fantastic: http://www.youtube.com/watch?v=BttI-y9VzXQ http://www.youtube.com/watch?v=BttI-y9VzXQ

moron4hire 16y ago

One of the points he makes is that VMs are obvious for the purpose of language interop. Seems like C and Lisp (and D, another language he mentions) have been interoperating with each other without a VM in common for quite a while.

Goladus 16y ago
But it's not a many to many mapping, that's the point he was making. Lots of languages interop with C, but you can't just import a ruby module into a python script very easily. In fact if you want to link to C libraries from D you have to modify the C header files to turn them into D modules.
- moron4hire 16y ago
  Sure, but it doesn't seem intractable to be able to design native languages to be able to interop. Lisp has been around since the late 50s, C since the 70s. The fact that they interop at all is kinda awesome.
- andrewf 16y ago
  I think Microsoft actually did a reasonable job with the "original" COM (ie OLE 2.0 - the stuff that let you put an Excel 95 spreadsheet into a Word 95 document). It had a bit too much bolted on, probably due to its origins in OLE, and I think "dynamic OLE" was a tragic mistake. And its reference counting approach was a lot more appropriate in the early 90s than it is today (where a compacting GC may be the smart choice, performance-wise). But for making calls between objects in C, C++, VB6, Delphi, and pretty much every other language that lived on the Windows platform, it worked really well.
  - limmeau 16y ago
    For a compacting GC, you'd have to hunt down a complete set of references from running programs written in unknown languages, with strange errors if the set is underapproximated even once. I'd rather have reference counting.
nickik 16y ago
It works but it is WAY HARD. Interact from ruby to common lisp. On a VM its easy a call from Jruby to clojure is not to hard. Its stupid to implement your own GC with every knew language and there are way more things that every C-Based language has to reimplement. Thats why it always took around 10 years until a language got really good and fast. Today you can have a language in 2 years with the biggest set of libs out there.