home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.os2.networking
- Path: sparky!uunet!noc.near.net!vaxeline.ftp.com!minimillian.ftp.com!backman
- From: backman@vaxeline.ftp.com (Larry Backman)
- Subject: Re: close-on-exec/IBM tcpip dev.kit.
- Message-ID: <921119101946@minimillian.ftp.com>
- Nntp-Software: PC/TCP NNTP
- Lines: 17
- Sender: root@vaxeline.ftp.com (vaxeline.ftp.com root account)
- Nntp-Posting-Host: minimillian.ftp.com
- Organization: FTP Software, Inc., Wakefield, MA
- References: <1992Nov10.082459.9024@kvamdata.no> <-159809996@hpopd.pwd.hp.com> <1992Nov16.202453.18247@ans.net>
- Date: Thu, 19 Nov 1992 10:19:46
-
- In article <1992Nov16.202453.18247@ans.net> db3l@ans.net (David Bolen) writes:
-
- >> The global sockets have advantages and disadvantages. It would
- >> probably have been nicer if they worked semantically just like file
- >> handles, but that wasn't possible in the first releases of IBM's
- >> TCP/IP (as its kernel was in user space). I've often wondering how
- >> well it would work to handle sockets with an IFS driver though, so
- >> they were actually file handles. The socket calls would create and
- >> access files in the \DEV\TCPIP tree or something (the IFS would be
- >> device IFS (\dev), rather than installing a new disk letter), and they
- >> would work just like file handles (ala Unix).
-
- Your not the only person who's often wondered about writing a
- socket.ifs.
-
-
- Larry
-
-