LLL,DIS: object file format

[Mike Prince]

On Mon, 31 Oct 1994, Francois-Rene Rideau wrote:

>   We will have to choose some object file format.
>   Why not having *it* be the solution ?
Things can get confusing from here.  I've proposed not having files.  Let 
me explain.  My first version of the microkernel wil use two disk 
directories as in and out buffers to other workspaces (microkernels).  
The in directory will have a number of files there, which will be treated 
as incoming messages.  The kernel when it is activated will read each of 
these "messages" in, and either crack it open into an agent or tool box 
(kernel doing the dirty work), or pass it to a higher level module for 
the cracking.  When you shut a kernel down, it simply bundles everything 
up (in this case only tool boxes, because all the agents will be in them) 
and puts them in the kernels IN directory.  That way when it is executed 
again everything comes up just where it left off.

Now this is different than the view a programmer may have of objects that 
are encapsulated within tool boxes.  We have two different perspectives 
to consider; the low level kernel view of whats going on, and the 
environment created by the language/programming system a person uses.  

>   To me, an object file can contain any system *object*, and not
> only code.

All this also underlines the need for our glossary (of which I've taken 
responsibility) and which is now becoming very needed.  By tomorrow it 
will be ready for discussion.  (But isn't tomorrow always a day away).  
OK. It will be done by the 1st of Novemeber.  Am I talking to myself?

Mike