[net.usoft.s] Cascade of S Bugs

naomi (03/07/83)

#N:ubc-medgen:3400009:000:2237
ubc-medgen!naomi    Mar  3 11:26:00 1983

Today, trying to be a good citizen, I ran the following:

nice -19 S < cmd > output &

(OK, I should have run the following:

nice -19 S BATCH cmd output &

but as it happens, exactly the same problem arises if you do it that way.)

In the first line of cmd, I try to save all the relevant data sets, in case
I mess them up in the subsequent commands.  Unfortunately, I made a
mistake in the save command, but

Bug 1: S, instead of quitting, continued to process the command file. You
       are supposed to be able to set the sourceexit command using options,
       but this doesn't work on our system.  

However, I didn't realize that S was still running, so I corrected the
error and reran the program. Unfortunately, following the
suggestion in section 9.1 of the manual, my datasets are prefixed by
directory names so

Bug 2: I could not save my data sets because I did not have the appropriate
       subdirectory under sdata.  (I guess this isn't really a bug, but that
       suggestion in the manual should go, since datasets prefixed by
       directory names do not work "properly" with several commands,
       including list.  For me, however, the benefits of using subdirectories 
       still outweigh the disadvantages.)

Nevertheless, after printing the error message, S continued to run, so I
then had two versions of S, both changing unbacked up my datasets, in the
background.  Checking on another background job, I discovered that S was
still running, so I brought one into the foreground, and  out.  Then I
brought the other one into the foreground, hit , and found myself in

Bug 3: an infinite loop of:
       Dumped
       jstkrl: tried to release < 0 items

Several screens of this message convinced me that I could not break my way
out, so I logged onto another terminal and 

Bug 4: I typed S KILLEM, which killed, not only my S job and all its
       subprocesses but also my stopped emacs job and all its subprocesses.
       Of course, it did not set emacs exit switch, so emacs protested
       (fortunately not too violently, when I tried to start it up again.)

Ah well, in future I will back up files interactively.

            Naomi Altman
!decvax!microsoft!uw-beaver!ubc-vision!ubc-medgen!naomi