$Id: README.Microsoft,v 1.2 1999/03/03 23:17:10 rossini Exp $

An Initial FAQ List for ESS problems under Microsoft Windows 95/98/NT.
======================================================================

Also see the file:

	README.SPLUS4WIN 

for more information and details.

* (RMH) system freeze and ctrl-m, a solution for some cases.

	Frequently, when the *shell* buffer freezes while running
	either COMMAND.COM or a DOS or Windows program initiated by
	command.com, it can be unfrozen by entering the key sequence
	C-q C-m RET into the frozen buffer.

	What I think is happening is the coding-system got confused.
	Command.com is waiting for the C-m C-l sequence and it only
	sees the C-l, which it knows is not the end of the line.  When
	we manually give the C-m, followed by RET, it is happy and
	completes parsing the line.

	For example, from a *shell* buffer running bash (with ps in
	the path):

	command.com
	ps
	C-q C-m RET
	exit
	C-q C-m RET


	command.com /c date
	RET
	C-q C-m RET

	command.com /c date
	C-q C-m RET

* (RMH) Comments:

	Note to ess-bugs and R developers.  To be broadcast to
	ess-help and S-news after several of you have checked this out
	and the files are moved to franz.  
	
	------------------------

	1. Error in ess-iw32.el in ESS 5.1.2

	In response to the discussion on this list, I tested ESS 5.1.2
	with R and discovered that as distributed it does not work
	with R.  The distributed version of the file ess-iw32.el
	bypassed a critical if-test.  I apologize for that error.

	There are 3 locations in ess-iw32.el where the two lines
	     nil)  ;;; temporary value.  Next line is what it should be.
             ;;;(default-value 'inferior-ess-ddeclient))
	must be replaced by the single line
	     (default-value 'inferior-ess-ddeclient))

	With this correction ESS 5.1.2 will work with Rterm.exe in an
	NTemacs instance where the shell is MS-DOS when the
	instructions in the file R/rw0632/README are followed.

	2. Improvements, currently in essd-r32.el.  To be absorbed
	   into essd-r.el.

	One of the design goals in ESS was to make it almost
	unnecessary for users to make modifications to the ess-*.el
	files.  So I revised essd-r.el to use the argument "--ess "
	for windows and the argument "--no-readline " for unix.  No
	user changes are required.

	M-x R

	When used with MS-DOS as the NTemacs shell, no changes are
	needed by the user.  The emacs command works correctly with
	either bash or unix as the shell in emacs.

	------------------------

	The first version of these changes are in a separate file
	essd-r32.el, which must be loaded into NTemacs after
	ess-site.el has been loaded.  After testing by others, it will
	be included into essd-r.el and distributed with the next
	update to ESS.


	A corrected version of ess-iw32.el and the first test version of
	essd-r32.el are on my website:
	  http://astro.ocis.temple.edu/~rmh/ess/

	The essd-r32.el files sets
	(setq-default inferior-R-program-name "c:/Progra~1/R/rw0632/bin/Rterm.exe")

	If this is not correct for your machine, you may need to
	change this one line.  This line will eventually appear in
	ess-site.el, so a user change here is permitted.

* (RMH) More comments about files and ideas:

	I have sent six files on my astro page

		 http://astro.ocis.temple.edu/~rmh/
	
	I attempted the cvs at franz and still don't have it figured
	out.  CVS.EXE [checkout aborted]: cannot connect to socket:
	Connection refused

	ess-iw32.el  correction.

	  The first version handled S+4, but not R.  The second version
	handled R, but not S+4.  This one handles both.  The problem
	was handling the (in my mind) broken ess-setq-vars-default
	usage in ess-inf.el.  In this version of essd-s+4 and ess-iw32
	I set the ddeclient variables outside of ess-customize-alist.
	This way I can distinguish between global and local values for
	tests.

	essd-s+4.el

	  Adjusted to match the change in ess-iw32.  The ddeclient variables
	are set outside of ess-customize-alist.

	msdos.el

	  Start an msdos shell inside an NTemacs that uses bash as its shell.
	It should have been easy, but had many parameters that needed
	to be set.  I am sending a copy of this to the Ntemacs mailing
	list.  I think this function should be merged into the Ntemacs
	distribution, but for the moment I am attributing it to us.

	msdos.el.cmts
	  comments to the NTemacs mailing list
	(AJR: merged with msdos.el, as a comment inside).

	essd-r32.el
	  Currently a separate file that depends on essd-r.el.  It should be
	merged into essd-r.el.  I originally intended to use M-x msdos
	and then run Rterm from there.  It turns out there is no
	advantage that way and it would be much harder to explain.
	There are still glitches in this interaction of R and ESS
	together, documented in the startup message when you run M-x R
	(after loading ess-site and essd-r32).  I think they are
	consequences of Rterm not being designed to run with bash.

	ess-bugs.cmts
	  A copy of this email message.
	(AJR: merged into README.Microsoft).