Re: CVS is the `released version'

[Ryan Yeske]

       In my quick test of package.el, I was able to install and run a
       package in 1/10th the time it would have taken to find, download and
       install it manually.

   Can you explain why this is so?  What are the jobs that you need
   to do without package.el, which package.el avoids?

I installed package.el.  Then I ran package-list-packages, was
presented with a list of available packages in a buffer.  I hit RET on
wtf, which is a package that looks up acronyms.  package.el downloaded
the code, and installed and evaluated autoloads.  I was then able to
simply do M-x wtf to play with the new code.

Had I done this manually, I would have had to locate via a web search
where to ftp wtf.tar.gz (or whatever form it would be in), download it
via ftp or http, save it to my personal elisp directory, untar it, and
read the INSTALL file, or the commentary in the elisp file to find out
what autoloads to add to my .emacs file.  I would never have bothered
trying out wtf.el.

       Having the package system install the file and setup autoloads in
       .emacs will not itself change emacs functionality, but will save the
       user a tremendous amount of tedious work.

   I can see how putting the autoloads in a suitable place would be a
   savings.

   That is something that could be done by a function to install certain
   Lisp code, which does not need that Lisp code to be a "package".

Perhaps we are talking about different definitions of package.  I am
not referring to package system in elisp, in the same sense that
common lisp has a package system, but simply a system for automating
the routine retrieval and installation steps for making emacs
extensions available.