Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!mcnc!rti-sel!sas!burnette
From: burnette@sas.UUCP
Newsgroups: comp.sys.amiga
Subject: Re: The resident command
Message-ID: <229@sas.UUCP>
Date: Sun, 23-Nov-86 12:23:09 EST
Article-I.D.: sas.229
Posted: Sun Nov 23 12:23:09 1986
Date-Received: Mon, 24-Nov-86 19:31:31 EST
References: <2062@well.UUCP>
Organization: SAS Institute Inc. Cary, NC
Lines: 31
Summary: The RUN command doesn't know about it. Thoughts on how to make resident C programs.



The resident cli seems to have one fatal flaw:  RUN doesn't know
about resident commands.  For example, if you enter
   resident md c/makedir
and then try
   run md
the md command will not be found.  This occurs even in execute files
started by the run command.  For example:
   run execute foo
where foo contains an 'md' command.

Other than that, resident commands work very nicely.  There is a noticable
speed improvement over executing the commands from ram: (instant versus
a fraction of a second).  Then of course there is the memory savings of not
having two copies of a program in memory.

Both the Lattice and Manx compilers support base register relative data
addressing now so it would seem possible for us to come up with a scheme for
resident commands that takes advantage of that feature.  When you make a
command resident the code and data sections could be loaded into memory.
Then, when someone executes the command, a copy of the data section could
be made; the bss part could be allocated and initialized; a base register
could be pointed to the top; and then the code that is already in memory
could be executed.  The bss section will need to be allocated at the end
of the data section, I think.

The only problem I see is that Lattice and Manx assume different base
registers for the data.  We either have to detect somehow which register
to use or get the two companies to agree on a standard register (perhaps
some register in between the two they use now :-).