On Sun, 21 Jan 2001 13:16:21 +0900, "Stephen J. Turnbull" said:
First off, I'm sorry, but I miswrote. What I really wanted to
know
was "suppose we document it. Then do you care?" The installed dump
file implies it's part of XEmacs by its name. If this were properly
documented in Info, FAQ, INSTALL, NEWS, and Installation, would you
still object to this file being installed in .../bin?
That would be much more acceptable. ;)
A case could be made that it belongs in .../share insteady, however. ;)
Nicholas and Martin evidently object because it's not an
executable
command. Is that the way you feel about it, or do you worry primarily
because it's undocumented?
It's close enough to a binary I can live with it being in .../bin. I
get torqued off when I find stuff in /usr/local and can't identify who/what
put it there. ;)
Are there _tools_ that sweep bin directories for non-executable
cruft
that would care, to anybody's knowledge? Is that a sufficiently good
idea that even if they _don't_ exist, someone should write one (and
probably will)?
No, because most people who are *really* concerned about keeping track
of /usr/local probably manage it by way of RPM or similar tool. I used to use
Depot, but on AIX have been building 'installp' images for anything that
goes into /usr/local (among other things, this guarantees that if I
uninstall it, all the files are removed.
This however raises another question - does the current pdump support
"build in ONE directory, do 'make install' into a staging tree, <magic
happens>, and the files are in /usr/local? This is an issue for
AFS, Depot, and installp at the very least - I'm not enough of an RPM
person to know if it's the case there too. In fact, this is probably
*more* crucial to me - a software package management system will *tell*
me what fileset a random file belongs to. But I need to be able to
package it first... ;)
/Valdis