"Linking" drivers to a binary kernel

[-m32]

Not only is it daunting, it is extremely involved and time consuming.

I've written a compiler & lexical/syntax analyser before. It is no small task.

It could take years to work out all of the semantics in getting such a thing to work for real-world applications. Also, what about language? Should one create a totally new language to accommodate requirements? Use an existing language? Extend an existing language?

Nah, that's beyond what I want to do. I'm already re-inventing one wheel

Thanks for the suggestion though.

I'm still investigating solutions for my particular implementation. Once I think of something brilliant, I'll come back and update this thread

[Combuster]

I've written a compiler & lexical/syntax analyser before. It is no small task.

How to make a java to native compiler in one day: Know haskell (proven possible, I spent less than 12 hours on that assignment )

*ducks and runs*

[fronty]

How to make a java to native compiler in one day: Know haskell (proven possible, I spent less than 12 hours on that assignment )

Do you have sources somewhere available? I'd like to see how well pure functional languages suit to compiler writing, I have written/participated writing compilers and interpreters only in procedural languages, if you don't count some small interpreter for subset of K&R C in Lisp. I guess something like recursive descent parsing is quite natural thing to write in functional languages.

[-m32]

How to make a java to native compiler in one day: Know haskell (proven possible, I spent less than 12 hours on that assignment )

*ducks and runs*

Smart @$$

[egos]

You can include in a driver the import table, which contains standard export indexes for used kernel functions and terminator at the end, and store its address as a global symbol like a driver entry point. Or driver can register this address (for resolving table entries) by calling special function through int interface or through function address parameter transferred to driver.