lkml.org 
[lkml]   [1998]   [Oct]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Directory name problem...


On Tue, 27 Oct 1998, Mike A. Harris wrote:

> Naming a file "core" may not be the smartest choice of a filename
> on systems that have complete idiots as sysadmin's, but it is
> a perfectly legitimate filename/dirname, and the linux source
> tree illustrates this point, as do other programs/files.
>
> One other point nobody has mentioned is that on many systems,
> non-unix filesystems are mounted such as msdos/vfat, floppies,
> ntfs, hpfs, nfs, etc...
>
> I just did a filefind, and found a couple dirs named "core" in my
> msdos partitions. This stuff is NOT backed up, and would be a
> pain to locate again.

And when something will coredump there you'll do what? Any
regular file named 'core' has a chance to be overwritten. You don't
need a script to destroy its contents. And IMO if you are keeping
anything that is worth preserving (a) on FAT filesystem and/or (b) NOT
backed up... you are asking for trouble.

> Point being that opinions on the naming of files as "core" make
> no useful contribution to the problem that people DO infact use
> files and dirs with such a name, and may not know anything at ALL
> about Unix, or corefiles. User's might be running a Win95 system
> served by a Linux Samba server, and decide to name a document
> "core" for some project that is being worked on or something. In
> this respect, an innocent user is CERTAINLY not "asking for it".

There is no such thing as innocent user ;-/ Moreover, if you need
to exclude some fs (or fs type) from the search you know what to, right?

[snip]
> What's more is that someone may have a *legitimate* "core" file
> that is in fact a real core file, that they ARE USING, or ARE
> WANTING TO ANALYZE with gdb. This could for example be the core
Than rename them.
> file of a program that cores once every 18 days or so...
> Certainly the person isn't going to be very pleased to wait
> another 18 days to get another corefile to examine. How then do
> you propose to remove THESE core files? Someone has proposed
> checking the atime on the file. What if joe programmer is on a
> month's holidays? A lot of other what-if's come up too.
Really. What if Joe Luser Programmer keeps 25M of coredumps
and doesn't bother himself to touch them?

> The proper solution is to implement the script properly, and
> tweak it to a particular site's preferences, perhaps ignoring
> /home when scanning.

Sure. Ignoring home is generally _not_ a good idea, simply because
your beloved average luser _never_ removes anything that could be safely
deleted (OK, if he removes _everything_...)

> You certainly cannot expect every computer user to not use "core"
> as a filename to make up for the short-sightedness of a retarded
> sysadmin.
OK, admin in question is no doubt idiot, but could you clarify
your, erm, opinion on Thompson and Ritchie? Any writable regular file
named 'core' may be overwritten by dump. It was so on v6 and v7. It _is_
so on Linux.

> And, Linus/Alan/Donald/whoever was responsible for
> naming the directory "core" are certainly not "asking for it".
Right. Name isn't too nice, but it's a directory, _not_ a regular
file.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:45    [W:0.271 / U:0.012 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site