[ESS] Installing, testing various ESS, does roxygen -> html work for you?

[Paul Johnson]

For the past two weeks or so, I've noticed that the ESS / Roxygen /
HTML preview does not "just work" anymore.  I notice this occasionally
going between machines, I have not yet detected a pattern.  Problem
might be in the platform, Emacs edition, ESS, or the roxygen2 R
package, I don't know. Its too bad because I've been using this as one
of the shining examples of why Emacs and ESS are helpful (my traveling
road show 'Emacs has no learning curve').  The symptom is that the web
browser doesn't pop open with the preview.

If Roxygen -> HTML preview does not work for you, I'd appreciate
knowing because I'm trying to find out what causes it to fail.  I've
been testing various ESS editions on Ubuntu and RedHat and have had a
series of non-reproducible errors. On Macintosh it works fine, but
there are only 2 test cases. On Thursday, during  a lab session, we
found failures on MS Windows. I did not realize the roxygen2 version
may be at fault, I'll find out next time I see those computers.

I don't expect you can help with that, but I want to ask for a little
clarification about installing ESS snapshots.  Mostly my questions are
about "which ESS gets used"?

I'm in Linux systems where there is generally an older ESS installed
in /usr/share/emacs24, either by an RPM or by a deb package from the
system.

Question 1.

Is this right: If ESS is installed in the system, there's no way to avoid it.

When ESS is installed by a deb or RPM package, it includes a launcher
that goes into the
site-start.d folder. It is not avoidable to the users.

In a test system, I remove the system ESS package.

The other questions I have today are about choosing between various
ESS folders in a given system, after removing the system-wide ESS.

Question 2.

If I have several versions of ESS installed in my user account, from
source or from Melpa, what determines whether Emacs uses them?

Here's why I wonder....

I got ESS from GIT, copy the whole directory into

/home/pauljohn/.emacs.d/ESS

I ran make in that directory. Then in the ~/.emacs.d/init.el file I
inserted this:

(add-to-list 'load-path "/home/pauljohn/.emacs.d/ESS/lisp/")
(load "ess-site")

After that, ESS is found, I am using that version.

Question 2A. Can ESS from Git live in peace with ESS from Melpa?

Here's why I wonder...

I've mentioned in here before that I admire the package
ess-smart-underscore. Everybody should! Its awesome.

In my ~/.emacs.d/init.el, I have the "Melpa stanza":

(when (>= emacs-major-version 24)
  (require 'package)
  (package-initialize)
  (add-to-list 'package-archives
           '("melpa" . "http://melpa.milkbox.net/packages/") t)
  )

After that, open Emacs and run "M-x package-list", and there is a
more-or-less obvious way to install package ess-smart-underscore. The
package installs into ~/.emacs.d/elpa.  In my init.el, I add

(require 'ess-smart-underscore)

Shortly after I did that on Thursday, then ESS quit working. Help did
not work, pretty much everything died.  Then I realized ESS was
updated inside Melpa. The Melpa ESS perhaps had a bug.  I can't say
what version it was, I deleted it (in anger :().

Today I decided to try to re-trace my steps. This init.el has the Git
ESS load bits before the Melpa stanza, and it seems like the Git ess
stays in control:

(add-to-list 'load-path "/home/pauljohn/.emacs.d/ESS/lisp/")
(load "ess-site")
(when (>= emacs-major-version 24)
  (require 'package)
  (package-initialize)
  (add-to-list 'package-archives
           '("melpa" . "http://melpa.milkbox.net/packages/") t)
  )

(require 'ess-site)
(require 'ess-smart-underscore)

This surprised me a little bit because I expected the one requested
LAST would be in control.  But it appears the first in is in control.
Same as problem with ESS from rpm or deb package.  They are locked in
before the user account is touched.

And, guess what after all of this installing, uninstalling, etc. Now
Roxygen -> HTML works flawlessly. Humbug.

-- 
Paul E. Johnson
Professor, Political Science        Director
1541 Lilac Lane, Room 504      Center for Research Methods
University of Kansas                 University of Kansas
http://pj.freefaculty.org              http://crmda.ku.edu