home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!decwrl!decwrl!simpact!mvb.saic.com!macro32
- From: JWILKINSON@HMCVAX.CLAREMONT.EDU (James Wilkinson)
- Newsgroups: vmsnet.internals
- Subject: I/O to terminal server ports.
- Message-ID: <01GMQJW2Y1OWAOKE4L@HMCVAX.CLAREMONT.EDU>
- Date: 24 Jul 92 08:22:00 GMT
- Organization: Macro32<==>Vmsnet.Internals Gateway
- Lines: 30
- X-Gateway-Source-Info: Mailing List
-
- I have encountered several programs lately which rely on certain terminal port
- behavior which is not emulated by terminal server ports. When doing single
- QIOs to a server port, for instance, the request will not complete as long as a
- printer is not hooked up to the port (DS90L+). With a normal terminal device,
- the request completes right away and the I/O packet goes to the bit bucket.
- Additionally, with a normal terminal device one has access to curent flow
- control information available in the device UCB, whereas such information is
- unavailable from LTDRIVER.
-
- I have in front of me a program which was designed around the ability to
- determine whether a QIO would complete in a timely fashion before the QIO was
- actually issued. For instance, if the printer were offline or out of paper,
- the device would be in an XOFFd state and no QIO issued. If the printer were
- not connected, the QIO would still complete. Naturally, this code is a shoe-in
- for RWAST when the printer is an LT device.
-
- Does anyone have any idea how to obtain similar functionality for the LT
- device? I'm looking for some way to query the terminal server about the port's
- readiness, such that if I go ahead and issue the QIO I know that it will
- complete soon thereafter. The program shouldn't have been coded in such a
- fashion in the first place, however changing over to a multi-threaded
- asynchronous mechanism will require a change in the entire design. The program
- drives multiple printers.
-
- JaW
- jwilkinson@hmcvax.claremont.edu
-
- ps. I suspect that LTDRIVER must keep flow control information somewhere,
- otherwise servers wouldn't work. However, the scenario with the printer
- disconnected is the real problem.
-