Great Circle Associates Majordomo-Workers
(September 1996)
 

Indexed By Date: [Previous] [Next] Indexed By Thread: [Previous] [Next]

Subject: Re: Make problem in 1.94b2?
From: Brent Chapman <Brent @ GreatCircle . COM>
Date: Mon, 23 Sep 1996 10:50:26 -0600
To: Evan Champion <evanc @ synapse . net>, Jason L Tibbitts III <tibbs @ hpc . uh . edu>
Cc: majordomo-workers @ greatcircle . com
In-reply-to: <32454062.44A1@synapse.net>
References: <ufaiv97fosz.fsf@sina.hpc.uh.edu> <1531.843341943@rubble.west.ora.com> <1929.843350820@rubble.west.ora.com><ufahgorf1oi.fsf@sina.hpc.uh.edu>

At 9:34 AM -0400 9/22/96, Evan Champion wrote:
>Jason L Tibbitts III wrote:
>> JP> Seems like a good guess.  Ditto for anyone who's tried to use "medit"
>> JP> to edit files (without doing a chown first), eh?
>>
>> Well, you have to run medit behind the wrapper anyway.  I don't quite see
>> how that applies.  (Does medit even work?  I've never used it myself.)

No you don't, if you're a member of the group that has write permission on
the files you're trying to edit (and the directory they're in, so that the
lock files can be created).

>I never use medit...  All I do is edit the files directly and when vi
>tells me that the file on disk is newer than the file I'm editing, I
>reload the file and keep editing.

I don't think that's a universal behavior of all the various versions of vi
out there.

>Suggestion: in addition to (in replacement of) shlock.pl locking, could
>Majordomo do (optionally) flock()/fcntl() locking as well?  What I would
>like to see is an entry in the majordomo.cf that goes something like:
>
># to use all three methods
>$lock_method = SHLOCK | FLOCK | FCNTL;
>
># to only use flock() locking
>$lock_method = FLOCK;
>
>Then just add an if statement like:
>
>if (($lock_method & FLOCK) == FLOCK) { do flock lock }
>if (($lock_method & FCNTL) == FCNTL) { do flock lock }
>if (($lock_method & SHLOCK) == SHLOCK) { do flock lock }
>
>That would make it even safer to not use medit if you can have Majordomo
>lock the same way as your editor.  In fact, you could then get rid of
>medit...

Problem is, this assumes that either the installer knows which locking
mechanisms are available on their system, and more importantly, which work!
I'm not exactly a novice sysadmin, but I can never remember from one OS
release to the next which works and which doesn't on one platform.

I chose shlock very carefully because it was something I was pretty sure
would work regardless of what platform it was running on.


-Brent

----------------------+----------------------------+------------------------
Brent Chapman         | Great Circle Associates    | 1057 West Dana Street
Brent@GreatCircle.COM | http://www.greatcircle.com | Mountain View, CA 94041
----------------------+----------------------------+------------------------
                   Internet Tutorials from the Experts!




Follow-Ups:
References:
Indexed By Date Previous: Re: resend's back channel to digest
From: Dave Wolfe <dwolfe@risc.sps.mot.com>
Next: Re: Make problem in 1.94b2?
From: Brent Chapman <Brent@GreatCircle.COM>
Indexed By Thread Previous: Re: Make problem in 1.94b2?
From: Jason L Tibbitts III <tibbs@hpc.uh.edu>
Next: Re: Make problem in 1.94b2?
From: Jason L Tibbitts III <tibbs@hpc.uh.edu>

Google
 
Search Internet Search www.greatcircle.com