On Sat, Jul 6, 2019 at 2:28 PM Paul Gilmartin < [email protected]> wrote:
> On Fri, 5 Jul 2019 22:45:12 +0000, Jesse 1 Robinson wrote: > > >... If what's entered is not a real NetView command, you get this: > > > > REPLY INVALID. REPLY WITH VALID NCCF SYSTEM OPERATOR COMMAND > > > >This can easily happen if the operator tries to reply to a different WTOR > but gets the number wrong. Often this WTOR remains in this state for weeks > or months. Until the next IPL. You never know what it didn't like because > the bad reply has rolled off immediately. So my final piece of advice. > Whatever mechanism you settle on, don't be like NetView. > > > On our test floor (perhaps too many) engineers had operator authority > (I was among them.) I remember watching another novice trying to deal > with such as: > 42 RESOURCE UNAVAILABLE. REPLY 'CANCEL' OR WAIT > R 42,WAIT > REPLY 'WAIT' INVALID > 43 RESOURCE UNAVAILABLE. REPLY 'CANCEL' OR WAIT > R 43,WAIT > ... > > This process doesn't necessarily terminate. It would have > been far better if it were: > 42 Resource unavailable. Wait or reply 'CANCEL' ... Keeping 'WAIT' out of the scope of recommended replies. > And using mixed case vs. upper case to differentiate instructions > to operator from suggested replies. I like that idea. But then again, we need to remember that this code is likely ancient, perhaps from the days before 3270 consoles. When UPPER CASE ruled the MVT+ world (it was a loud, angry time?). IIRC, there were some OS consoles which would automatically upper case for display, but there were others that would simply print a blank. Which is probably why these messages are in upper case only. > And one of my pet gripes is that the reply number is incremented > when the reply is rejected. It would be far more intuitive if it allowed > the operator to retry with *the*same* reply number. I know; WAD. > But the design is wrong in not considering that human factor. > As far as I know, the WTOR processing itself does not in any way validate the data entered as the reply. It only validates the length of the data. The code which issues the WTOR is what is validating the input. And, if unacceptable (in this case), simply does another WTOR. Which is why the reply number is incremented. The WTORs are completely separate so far as the system is concerned. It might be interesting if WTOR had a "reissue" option which would reuse the same reply number. But that would require more bookkeeping. Or instead, have WTOR have two new options: "return reply id" and "desired reply id" so that the first time, you could get the generated reply id, and then request that value be used as the reply id for another WTOR if it is available. A third optional option might be to have an exhaustive list of acceptable replies and have WTOR itself validate the data entered. But I think this is really off-the-wall and beyond the scope of what WTOR should be doing. Personal opinion. > > -- gil > > -- Money is the root of all evil. Evil is the root of all money. With that in mind, money is made by the government ... Maranatha! <>< John McKown ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
