bug#14064: info.info page has conflict with info package

[Eli Zaretskii <eliz@gnu.org>]

> From: Glenn Morris <rgm@gnu.org>
> Cc: 14064@debbugs.gnu.org
> Date: Mon, 01 Apr 2013 11:29:50 -0400
> 
> Eli Zaretskii wrote:
> 
> >> So AFAICS there is nothing clever the Emacs install rule can do
> >> here. As I said, I'd just make the rpm spec file unconditionally not
> >> install Emacs's info.info.
> >
> > How is an rpm more clever than Emacs?  What do they do that Emacs's
> > installation procedure cannot?
> 
> I won't explain this very well, but:
> 
> You configure+build+install once on one machine, making a binary rpm
> that other people install on their machines. When they do so, they don't
> run any of the Emacs Makefile pieces, the rpm just unpacks itself (like
> a tar file, I guess, but smarter, registering what file gets installed
> where, etc). It can choose to do various clever things at installation
> time, but these things are controlled by the rpm's specification file,
> not by any part of the Emacs normal installation rules.
> 
> AFAICS, there is literally nothing smart the Emacs Makefiles can do
> about this issue, because they are not invoked at the relevant point.

That's true, but why do we bother about Emacs being installed from an
rpm?  That's _really_ the job of whoever prepares the rpm.  They have
the same problem with GDB, for example, which installs libbfd.

I thought we are talking about "make install", where we do control
things.  Does it make sense to refrain from overwriting at that time?