UNIX for Dummies Questions & Answers

I'd accept one decade, but not two decades.
As usual jlliagre there are so many versions of unix that you have not used that you get an impression of standardisation - but only within your own experience.

Getting back to our thread orginator's query:- ...I think that you have moved on from fuser into file testing logic. What you have coded says:

1. Capture PIDs into /tmp/active.log
2. If /tmp/active.log exists and is empty AND /tmp/file2 exists and is empty, then write a message and exit, otherwise rename /tmp/file2 to /tmp/file2.timestamp

Moving on, you are renaming /tmp/file2 to /tmp/file2.timestamp every minute too. This may cause conflict just on the bad luck of timing anyway, but what you are seeing is that you fail the condition when /tmp/file2 does not exist. You then try to continue in the else clause and rename a file that does not exist, hence the error being displayed.

If you can describe the logic you are trying to write into your code, someone may be able to help you get the various tests and flow correct.




Robin
Liverpool/Blackburn
UK

fuser was definitely part of SVR4.0 released in 1989. It was formally specified in SVID 4 (1995) and later made mandatory for Unix98 systems.
I'm afraid you are confusing releases/distributions and implementations. You seem to believe every other Unix or Unix like release is a rewrite from scratch of everything. That's usually not the case, especially for basic Unix commands like fuser which are often based on SystemV source code on most commercial Unix systems released the last decade (and earlier). They can of course have extensions and cosmetic variations but they usually have very few behavior differences as introducing them would break customers scripts which is obviously something editors want to avoid. Two main branches of Unix like system that share few or no code with System V are the BSD family and Gnu based OSes, mainly Linuxes. BSDs had (originally) no fuser and Gnu/Linux has one which has the same useful split output behavior than System V. That's the reason why I was curious to know what Unix OS you were referring to when talking about discrepancy. My only guess is you were using some sort of stripped off / limited feature reimplementation like busybox.

@jlliagre
I am familiar with too many versions of unix (many mainstream and many others which fell by the wayside) and until we know what Operating System the O/P has I err on the side of caution. There have been recent posts for ancient SCO unix and HP-UX 10.0 for example.

I have a large collection of useful scripts which I have ported on numerous occasions with varying degrees of success. One runs "fuser" then looks up the actual process with "ps". Not suprisingly this script has many exceptions (expecially with the "ps" bit). Judging from my comment line in 1999 an anomalous "fuser" was in Dynix though I didn't record the exact version.

I have in the distant past worked concurrenly on early SunOS (Berkeley) and unix SV R2 & R3 as well as umpeen non-unix contenders - some of which were better but more expensive.
I always treat BusyBox as an exception. Similarly Xenix and SCO (unless we know the exact version).

Anyway, I think we have cracked the original problem about extracting the PIDs from "fuser".