[net.unix-wizards] Review of New Eunice Implementation

ARPAVAX:mo (10/08/82)

******************************************************************
	For research use only - Any reproduction or
	redistribution of these comments requires explicit
	permission of the author, and such permission is NOT
	implied by this notice.
******************************************************************

The following is personal opinion, and in no way reflects any
official view by anyone, anywhere.

Over the past several weeks, I have been using Dave Kashtan's Eunice
development machine, evaluating the latest version of Eunice.
I must say at the outset that I am quite favorably impressed.

My biggest problems with older versions of Eunice were lack
of filename mapping and robustness.  Both of these concerns have been
addressed in quite satisfactory ways.  The new filename mapping
code seems to work quite well and is extremely bullet-proof.
Some premeditated attempts to fool it were handled properly
and as the ultimate test, Dave is running UUCP without hacking
filenames!  (An acid test if there ever was one.)  The mapping code
handles arbitrary Unix names, complete with embedded blanks,
control characters, and names like "aA_.aA_.aA_." without
any loss of information.

As for the robustness issue, it has been rendered moot.  Most of
Eunice is now implemented in C and at no time did Eunice 
do something that you have to know VMS to understand.
As with any piece of software, there are probably a few
rough edges hidden in obscure places, but none of them
snagged me. (An aside: Dave has developed programming
schemes for interfacing with VMS facilities from C in
clean and simple ways.  This could be of great benefit
to sites doing VMS-specific programming in C.)

As for 4.1BSD-specific facilities,
the new version supports the "job control" facilities,
which make life much simpler.  They seem to be robust and work well.

I have only one complaint about the version of Eunice I have
been using - "stty erase ^H" doesn't work, but that is something
Dave can't really do anything about.  Until DEC decides users
deserve a choice of control characters, I don't know what
can be done easily. (But finger-macros from years of Unix use
just don't switch off instantly.) 

All in all, using Dave's new system was a pleasant experience.
As I said before, there are probably rough edges which will appear
only after heavy use by many Unix users, but Eunice appears to be
a useful, viable product which should be considered as one of several
possibilities by organizations looking for VMS productivity tools.

	-Mike