HP3000-L Archives

June 1996, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jeff Kell <[log in to unmask]>
Reply To:
Jeff Kell <[log in to unmask]>
Date:
Sat, 8 Jun 1996 02:08:48 EDT
Content-Type:
text/plain
Parts/Attachments:
text/plain (61 lines)
On Fri, 7 Jun 1996 12:08:00 P Ken Sletten b894 c332 x62525 said:
>Jeff Kell and anonymous after me:
>
>>>A question came up during a local discussion here
>>>at NUWC:  Is there (or will there be) any easy way to
>>>re-direct $STDIN and $STDLIST to an open socket ??
>     [...snip...]
>>If telnet integration was at the file system level, great.  If instead they
>>tweaked it at the low level like VTSERVER then your applications are still
>>as in the dark as they were to start with.
>
>Also, in another private response I got the following:
>
>>The file numbers are (currently) off, but there is(/was) a project to
>>realign those for 5.5.    [....SNIP....]
 
The "file numbers" issue is so that $stdin/$stdlist are consistent with the
Posix counterparts in the shell, which through 5.0 they are not.  To back up
to the original issue (cgi programming compatibility for "straight" MPE
apps) we only need $stdin/$stdlist redirected to sockets.  You can indeed
invoke shell scripts or Posix C programs as cgi processes on 5.0 and it will
work as expected.  It won't work with an MPE-compiled C program or any other
MPE-domain program.  $stdin/$stdlist don't talk to sockets.
 
>I hereby take it upon myself to "lift" Jeff's response and the
>other, tack it on to part of mine, and forward all to the Chair
>of SIGMPE (a.k.a. Tony).
 
My reply borders between technical (but not detailed enough) and general
(perhaps too detailed for management) but the crux of the matter remains.
Only Posix can talk to sockets in a "redirected stdin/stdlist" mode on 5.0
and even then not always -- only when dealt with as sockets.  Some code
relies on the socket/file descriptor compatibility as well; I found this
over a year ago trying to port the FreeBSD "finger" code.
 
>MPE Roundtable question.
>For HP World - Anaheim, August 1996:
>
>We need to be able to use file equations to easily
>re-direct output to a socket.  I believe that 5.5 realigns
>internal file numbers for CGI compatibility, but this by
>itself does not necessarily resolve all the issues
>involved in linking the file system to sockets.  What
>else needs needs to be done to allow a file descriptor
>and a socket descriptor to be freely interchanged ??
>And whatever that is, does CSY have any plans to do
>it ??
 
You don't necessarily need ":file" ability, since a socket must be bound
before it's a viable "file".  The principal need is to enable a process
operating with stdin/stdout on a socket to be able to invoke MPE code and
have the associated $stdin/$stdlist redirected to the parent's stdin/stdout
socket.  Again, I'm being too vague for technical readers, maybe too
technical for general readers.  It's a subtle issue, to say the least, but
one that keeps being swept under the rug.
 
It also isn't just a CGI issue; it "could" enable the shell and the CI to
co-operate, among other things.
 
Jeff Kell <[log in to unmask]>

ATOM RSS1 RSS2