NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/unix/aix/9564

Internet Message Format | 1992-09-10 | 3.6 KB

Path: sparky!uunet!haven.umd.edu!darwin.sura.net!spool.mu.edu!sdd.hp.com!cs.utexas.edu!ut-emx!ibmchs!auschs!awdprime.austin.ibm.com!konopik.austin.ibm.com!konopik
From: [e-mail] (Brad Konopik)
Newsgroups: comp.unix.aix
Subject: Re: 3.2 BSD Line discipline...
Message-ID: <[e-mail]>
Date: 7 Sep 92 19:09:39 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail] (USENET News)
Organization: IBM AIX Porting Center, Austin
Lines: 63

In article <[e-mail]> [e-mail] (Wolfgang Ksoll) writes:
>Brad,
>I am tired of hearing complicated stories about the serial port.
>For more than a year we hear things like:
>  [sample incantations deleted]
>I found that it is easier to add communication over ethernet, token ring,
>fddi,serial optical link, than to add a recent modem from the market place.

That's because we did such a good job on the pty driver :-)

>Even to add X.25 facilities (X.29, TCP/IP, OSI, SNA) is easier.

The X.25 stuff was NOT easier last time I tried it !

>Even to debug third party ISDN-drivers is easier.
>
>I think it is not the right way to tell us how bugs have to be reported
>through big blues channels. I think it is time that someone sits down and
>redesigns the device driver for the serial ports so that it is useful
>not only for terminals and printers but also for modern communication
>equipment. She should be ready some ten weeks before 3.3 is released so that
>some kind of quality assurance could be done.
>
>What do you think about it?
>
>Cheers,
>Wolfgang Ksoll           [e-mail]
>DATA SERVICE Berlin, Ansbacher Strasse 16, W-1000 Berlin 30, Germany
>Phone: [phone] - 32 FAX: - 39

I agree that the serial stuff has a few short comings, mainly where users
want non-standard function.  I like stuff to be intuitive...ours isn't, but
neither was the SUN or AT&T 5.4 boxes I tried.  Unfortunately, we DO have
to carry along the past 20 years of UNIX-development baggage...  I also
agree that some better quality assurance could be done.

What size staff do you think it would take to handle:
  - incessant flow of customer reported problems (about 3 of the 15
    reported each week are NOT user errors or how-to questions).
  - architecture and development of the device driver (for terminals, printers,
    modems, muxes, faxes, psuedo of X and network, etc.. etc..).
  - architecture and development of a friendly, intuitive user interface.
  - acquisition and execution of enough test cases for each of the areas that
    tty affects to ensure superior quality.
  - perform compatibility testing with other Un*xes.
  - technical writing to document the drivers, commands, libraries, and
    peripheral setup for this.

I'd like the staff to be composed of intelligent, full-timers with a decent
Unix background (not waves of contractors or temporary help that I have
to retrain every 6 months).

We've been trying to do it with two (sometimes one) dedicated bodies since
1987.  The same person supports(ed) multiple levels/platforms at the same
time (2 releases of RT, and RISC 3.1 and 3.2).  I used to average about
100 hours of overtime every month when I was the tty guy.  Now I just
coach from the sidelines in my "spare" time.

--
  tcpnet: [e-mail]  |  Brad Konopik
 ibmvnet: KONOPIK at AUSTIN               |  IBM AIX Porting Center, Austin
internet: konopik.austin.ibm.com![e-mail]
   uunet: ..!uunet!ibmsupt!ibmpa!konopik.austin.ibm.com!konopik