Re: CVS is the `released version'

[Tom Tromey]

>>>>> "Eli" == Eli Zaretskii <address@hidden> writes:

Eli> There are problems with this that no package system can ever solve: a
Eli> package that is released asynchronously from Emacs runs a high risk to
Eli> become broken by changes in Emacs.  Packages that are released
Eli> together with Emacs are generally coherent with the Emacs core, by
Eli> contrast.

Yes, that is true for some subset of packages.  But, it is by no means
universally true.

And, package.el makes an attempt to solve a related problem
experienced by users (me, at the very least): when an external package
is included in Emacs, package.el knows to disable the privately
installed copy in favor of the newer version included in Emacs.

Eli> As long as this is a real problem (and I personally don't see how it
Eli> can be solved, given the high rate of changes in core code), this
Eli> ``biggest benefit'' is actually a myth, IMHO.

I think you are drawing an overly general conclusion here.  A package
manager does not have to be perfect and solve every conceivable case
in order to be better than the status quo.  It merely has to provide
some tangible benefit to users.  Not solving every case does not
entirely eliminate the benefits of this application.

And, anyway, when a new Emacs is released, package maintainers can
update their packages.

Tom