Sigh...I'll try again. The first one seemed to get randomly formatted.
Re:
Glenn writes:
> Ron Horner writes:
> > Wouldn't it be easier to write a program that uses HPVOLINFO???
> That depends on your definition of "easier."
> It has been shown that productivity is improved when writing
> in a scripting language. Chances are, given the choice between
> writing a short CI script vs. a short compiled program, even
> today you choose the CI script.
Although you can probably get an apparently working result quicker via
scripting (and I did), you'll get a better, more reliable, more efficient
program if you write one using HPVOLINFO. What people forget about scripting
solutions is that they're generally not good at handling problems.
The script:
> #!/bin/sh
> callci "DSTAT ALL" |
> awk '$2 ~ /MASTER/ {x = $NF;
> sub("^\(?","/SYS/PUB/VOLUTIL \"showset ",x);
> sub("-[0-9]+\)?$"," classes\"",x);
> print x; }' >/tmp/vcl.$$
> . /tmp/vcl.$$
> rm /tmp/vcl.$$
relies on DSTAT ALL output, plus VOLUTIL output, plus the ability to create
files in /tmp ... all of which could change and/or have problems.
Heck, simply running VOLUTIL from the shell is fraught with problems... I
tried it, and ran into show stopping bugs.
Here's one example...in the following, note that the attempt to run VOLUTIL
and get the output into a file failed utterly, with no warning (on our 6.0
system):
/SIELER/HFS: /SYS/PUB/VOLUTIL "showset users;info=classes"
Mirvutil A.01.00, (C) Hewlett-Packard Co., 1991. All Rights Reserved.
volutil: showset users;info=classes
Volume-class name: Index:
----------------- -----
DISC 1
/SIELER/HFS: /SYS/PUB/VOLUTIL "showset users;info=classes" > foo
/SIELER/HFS: cat foo
/SIELER/HFS: ll foo
-rw-rw---- 1 STAN.SIELER SIELER 0 Jan 12 10:39 foo
Also, in some cases (which I saw when developing my first script, but I can't
reproduce now), VOLUTIL seems to get confused by the input parameter (INFO
gets corrupted?). In other cases, the shell reports a problem with "waitany"
(?).
In short, think of scripting like a 4GL ... yes, you can usually get an answer
faster ... but it isn't as robust or efficient as using a 3GL.
Stan Sieler [log in to unmask]
www.allegro.com/sieler/wanted/index.html www.allegro.com/sieler
Stan Sieler [log in to unmask]
www.allegro.com/sieler/wanted/index.html www.allegro.com/sieler
|