Xref: utzoo comp.emacs:4516 comp.os.vms:9687
Path: utzoo!yunexus!geac!syntron!jtsv16!uunet!dalcs!forceten
From: forceten@dalcs.UUCP (Safeguard Business Systems)
Newsgroups: comp.emacs,comp.os.vms
Subject: DEC-Shell bizarreness or bizarre Emacs?
Message-ID: <3065@dalcs.UUCP>
Date: 28 Oct 88 23:10:37 GMT
Article-I.D.: dalcs.3065
Organization: Math, Stats & CS, Dalhousie University, Halifax, NS, Canada
Lines: 59

][b
	This is not directly related to my previous posting about
    VMS/Emacs bizarreness.

	Emacs works entirely differently when invoked under the Death Shell.
    The terminal control stuff that doesn't work when Emacs is invoked
    from DCL works fine if Emacs is invoked from DEC-Shell. Unfortunately,
    the things that worked fine from DCL don;t work from Shell.

	For example, Emacs always thinks its starting directory is a
    particular starting directory in my login directory, no matter where
    in the file system my "DEFAULT" directory was set.  It also starts
    understanding a wierd hybrid of UNIX/DCL syntax for file specifications,
    such as /disk3/csar4250/csar0003[.utils]filename.  Directory name
    expansion almost works sometimes.  Why does the same program work
    differently under the Death Shell.

	A second, more exasperating problem, is that all of a sudden various
    Emacs packages stop working from the Shell!  For example, c-auto-newline
    no longer works.  My initialization file is being read normally, the
    c-mode-hook is being called as expected, and various variable are being
    set just as I expected.  The difference is that Emacs no longer seems
    to care about the values of these variables anymore.

	From the Death Shell emacs is being invoked by:

	Pathname/emacs -map Pathname/emacs.dump

	"-map" is an argument to emacs, not the DCL arcana that some might
    be familiar with, and IS being passed to emacs.

	NOTE to anyone from DEC involved with Shell that might be reading this:

	The DEC Shell (version 2) is a nice idea, but until mucho work is
    done, it will be the Death Shell to those who use it.  You might start
    by overriding the default CPU mis-allocation strategy of handing over
    1/2 of the available CPU time quota to every shell script that is executed.
    I think it does this to all processes.  An iterative shell script runs
    out of CPU time very quickly.

	Also, a null device is ABSOLUTELY NECESSARY.

	The redirection operators should create Stream-LF files, not variable
    record length ones.  The documentation claims that the package is a
    response to users who want to run UNIX packages on VMS.  Too bad most
    UNIX packages are written in C, and your C compiler, by default, uses
    the more versatile Stream-LF file organization.

	Not including a version of make is silly. If MMS is so great that it
    is worth the bucks DEC charges for it, it has nothing to fear from 'make'.

	DEC-Shell's advantage: It sure beats DCL for everyday use. So does
    MS-DOS, CP/M, and the later versions of NOS CCL.

||!][b-- 
Neil S. Erskine			MT&T - (902) 453-4915
Safeguard Business Systems	USENET { garfield, watmath, utzoo!utai,
6080 Young St, Suite 901.		 seismo, uunet } !dalcs!force10!erskine
Halifax, N.S. B3L-4H9