MIKE ROSS wrote in a message to Roy J. Tellason:

 MR> "Roy J. Tellason" wrote to "Greg Mayman" (16
Dec 02  20:06:09)  ---
 MR> on the topic of "THOSE OLD EXPENSIVE [1/2]"

 GM> I don't know why the 80xx software is so big. Does it include
 GM> workspace in the file, or what?

 RJT> I never got into it enough to figure that out.  But you're loading
 RJT> segment registers as well as the usual ones,  depending on which
 RJT> memory model you're using.

 MR> I know the answer to this one. It has to do with the compiler
 MR> libraries getting filled with all of kinds swiss army knife like
 MR> routines. Say one program uses but 1% of the library but it 
 MR> compiles the other 99% just the same because the required routines 
 MR> are so deeply embedded they can't be compiled alone anymore. For 
 MR> example you can't just print "Hello world" without first having 
 MR> called a few zillion odd subroutines which have nothing to do with 
 MR> illuminating spots on your phosphor. 

Especially if you use printf() rather than puts(),  as the former links all
sorts of stuff in,  like floating point even if it's not used anywhere else
in your program.  Yeah,  the granularity of one's library is definitely a
significant factor there...

 MR> These would for example be memory management, exception 
 MR> supervisory, break vectors, virtual machine, emm, hdd, blah... it 
 MR> goes on ad nauseum. Go and explain to me why an executing program 
 MR> which doesn't do any disk I/O has to access the hard disk at some 
 MR> point or other, it's not a virus, and there's plenty of ram so the 
 MR> virtual memory is turned off?! #$%{at}^!!! 

Depends on the tools you're using,  some stuff was way better at this than others...

---