🚀 go-pugleaf

>From Goldberg@RUTGERS Sun Nov  1 23:29:42 1981
I want to draw a distinction between program paging in general
and file mapping within an editor or other specialized system.
I do not propose that paging should be eliminated from the
operating system.  However, it is perfectly reasonable for an editor or
(other) data base system to map the file itself if the operating 
system does not provide this facility.  

In the case of a global operation on the file, the editor usually 
knows it will need the file pages in sequential order.  In this case, 
it can start to fetch the required file buffer pages in advance of 
when they are referenced its the search code.  If limited virtual 
address space and physical memory were not a problem, there would be 
no need to do so, but in the real world they usually are a problem.

As for paging the program code itself, again it is easiest and nicest
if the operating system can take care of this for you but not totally
impossible to achieve through overlays in certain special cases.

Before this discussion goes any further on this mailing list, let me
state that I am not interested in a "religious" discussion of which
systems, editors, are best, easiest to program on, etc.  Nor am I 
claiming that one can achieve the same performance from a micro with a 
16 bit address space as one gets from a DEC20, or any other such 
nonsense.  

My original message was in response to a statement that a 
PDP11-class processor is insufficient to support a reasonable text 
editor.  I take "reasonable" to mean a video editor that operates on 
files of arbitrary size with a well designed set of local editing 
commands plus global search, replacement, and copying, and which can 
search a file about as fast as the fastest disk read operation on a 
hard disk.  

My claim was simply that a PDP11-class processor can support such an 
editor if it is well implemented, not that such an editor can be 
implemented on a small machine as easily as it can on larger machines.
-------