NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/os/vms/18227

Internet Message Format | 1992-11-19 | 2.6 KB

Path: sparky!uunet!zaphod.mps.ohio-state.edu!darwin.sura.net!uvaarpa!concert!ais.com!bruce
From: [e-mail] (Bruce C. Wright)
Newsgroups: comp.os.vms
Subject: Re: The problem with $CREMBX System_Service
Message-ID: <[e-mail]>
Date: 19 Nov 92 09:32:26 GMT
References: <[e-mail]>,<[e-mail]> <[e-mail]>
Distribution: world
Organization: Applied Information Systems, Chapel Hill, NC
Lines: 39

In article <[e-mail]>, [e-mail] (Carl J Lydick) writes:
> In article <[e-mail]>, [e-mail] (Paul S. Winalski) writes:
> =
> =The DCL command interpreter uses RMS to read its input file.  RMS is a
> =record-oriented I/O system.   [...]
>
> The fault is not in RMS.  MAILBOXES ARE RECORD-ORIENTED DEVICES.  You can't
> read a single character from a mailbox (well, you CAN, by using $QIO[W], but
> the only character you can read singly is the first character of a record,
> and then you lose the rest of the record).
>
> =The solution is to buffer characters inside your program until you read a
> =CR from the terminal, then write the buffered string (not including the CR)
> =to the mailbox using one $QIO.
>
> Or have the application that's reading the mailbox buffer what it reads until
> it sees a CR in the data its reading from the mailbox.  Both techniques are
> equally effective, and I'd choose the one for which obtaining and modifying
> the source code is easier.

Another solution, which may be more like what the original author was
looking for, would be to use the pseudoterminal driver (either the FTA
driver or the PY/TW drivers depending on VMS version and/or convenience).
It wasn't entirely clear to me whether the applications were all newly-
written applications or not;  and in any event, from the point of view
of the consumer application, a pseudo-terminal is (nearly) indistinguish-
able from a real terminal, so even if it isn't newly-written it would
probably be very easy to make it run on a pseudo-terminal.

You can often use mailboxes for this sort of thing, but as Carl points
out there are problems because of the record-oriented nature of mailbox
I/O.  But the pseudoterminal drivers are true stream-oriented devices
that allow you to do single-character I/O in the manner that the original
author was expecting the mailbox driver to operate.  They're both useful
devices, but in general mailboxes are more useful for sending highly
structured messages, and pseudoterminals are more useful for capturing
terminal-like I/O streams.

Bruce C. Wright