NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/unix/bsd/5424

Text File | 1992-09-09 | 4.9 KB | 101 lines

Newsgroups: comp.unix.bsd
Path: sparky!uunet!spool.mu.edu!caen!hellgate.utah.edu!fcom.cc.utah.edu!cs.weber.edu!terry
From: [e-mail] (A Wizard of Earth C)
Subject: Re: AT&T Long Distance Boycott (was: BNR2SS, Mach, and The Lawsuit)
Message-ID: <[e-mail]>
Sender: [e-mail]
Organization: Weber State University  (Ogden, UT)
References: <[e-mail]> <[e-mail]> <[e-mail]>
Date: Thu, 10 Sep 92 01:46:18 GMT
Lines: 89

In article <[e-mail]> [e-mail] (Sean Eric Fagan) writes:
>In article <[e-mail]> [e-mail] (Terry Lambert) writes:
>>    What about the termcap file itself?
>Hate to tell you this, but I have seen applications that came with their own
>termcap (or termcap equivalent) file.  Yes, the whole thing.
>
>Next complaint?

Carrying around a termcap file rather than using native facilities, for\
one.

>>    VMS doesn't support the concept of SUID/SGID necessary from most
>>things, like score files in games.  Instead, you have to install the image
>>with priveledges (ie: I can't make an SUID "terry" program; only an SUID
>>SYSTEM with some priveledges).
>
>Gosh.  Then I guess the fact that such things were in the last VMS I checked
>was my imagination, huh?  Yes, amazing as it seems, the POSIX work done to
>VMS is real, and it works, and it does make it look a lot more like a UNIX
>system.

    SUID?  And this is without a special security dispensation?  The VMS
equivalent of this (SET UIC) has always required CMKRNL in this past; if it's
as a priviledged library module, this has generally required some priviledge
on the part of the user to link.  What precise version of VMS are you
claiming this for?

>>    The total lack of a CBREAK mode disallows single character I/O,
>>unless you are willing to call SYS$QIO() directly.  Raw mode requires a
>>QIO with IO$_SENSEMODE/IO$_SETMODE.
>
>Here's another concept you and Peter seem to be missing:  wrapper libraries.
>The VMS POSIX stuff has termios, which allows single character I/O the same
>was as on a SysV system (well, kinda).
              ^^^^^^^^^^^
I have a problem with this.

>[ ... nice example of how record I/O breaks tell/seek operations not on
>      record boundries, and the tell return value (record number) is not
>      correctly advanced after the carriage control is faked by RMS ... ]
>
>Works fine.  Oh?  Don't you use the correct port of emacs?

I kind of doubt that it "works fine".  Still fails miserably for me on a
*very* recent version of VMS.  The point was, as you editted it to appear,
the ability to create record oritented files, but, instead, the ability
to use record oriented files as if they were stream oriented files.  I
haven't even gotten into the fact that lstat() doesn't return real
information, like POSIX says it should, about record oriented files, or
the fact that, since there is no file system cache on VMS, things are
actually updated, rather than "marked for update", as POSIX mandates.  The
first, of course, will show up on a POSIX test suite; the second,
unfortunately, does not (if it did, it would force in-kernel caching to
be implemented, which would be a good thing -- but then again, maybe not,
since the "sync" keyword is already taken in VMS).

>Note that MMDF won't work on a lot of systems because it is
>highly UNIX dependent, and if you claim a POSIX-dependent application isn't
>going to work on a non-UNIX system, then I don't see how you can think that
>something like MMDF will.  (Networking?  IPC?  fork() and exec()?  setuid()?
>seteuid()?  Huh?  What are those?)

All of those are *very good* reasons POSIX compliance isn't enough to
support *real* applications (thanks for the ammunition, Sean)... but mmdf
isn't one of them.  When the Army developed the original mmdf at Fort
Huachuka, the original platforms were where Intel OpenNET ran (the first
commercial OSI network implementation), which were Intel 310's, Intel 320's,
and VMS.  OpenNET was also used with an INT 5C redirector from DOS, but
since DOS isn't multitasking, there was no real reason to port it there (piss
poor mail facilities, don'tcha know).


Like I said before, POSIX is a good first step.  I still maintain it isn't
enough, now that system libraries have pretty much been standardized, and
they should be part of "Open VMS" as well.  Curses was one example, termcap
is another, and there are many, many others that come to mind.


                    Terry Lambert
                    [e-mail]
                    [e-mail]
---
Any opinions in this posting are my own and not those of my present
or previous employers.
--
-------------------------------------------------------------------------------
                                                       [e-mail]
 "I have an 8 user poetic license" - me
-------------------------------------------------------------------------------