[ESS] Release of ESS 5.3.5

[Phillip Lord]

>>>>> "MM" == Martin Maechler <maechler at stat.math.ethz.ch> writes:

  PL> As I use bash, this always returns....

  PL> bash: c:/Program: No such file or directory\n

  MM> aha! You *are* using Windoze

Yes. Sorry I had assumed that this was implicit, but I had misread Stevens
email. He said "especially if you are windoze". 


  PL> and it all goes wrong from there.


  PL> Would checking c:/Program Files/R/ for directories not work reasonably?

  MM> No, as we just heard on ESS-help, R may well be installed in a different
  MM> place.

  PL> If

  PL> c:/Program Files/R/R-x.x.x/bin/R.exe

  PL> exists then you can snarf the version from the R-x.x.x without messing
  PL> around with processes.

  MM> Yes, but what if things are in d:/R/ instead -- a very reasonable
  MM> choice?

Well, finding all the versions of R is going to be heuristic, so there is
nothing to stop you from doing both. I would suggest the following:

1) Check a variable set by the user -- they can always override things. 
2) Check c:/Program Files as this is the canonical and default location for
   storing programs on windows. 
3) Check the path. 

Then the user with d:/R could either set a variable, or add R to the path,
while the user who just accepted the defaults would have it work
automatically. 

In general, these days I try and stick with system defaults where ever
possible, at least for stuff not under my control. Even miktex and emacs are
in c:/Program Files now. 



  PL> BTW, just install ESS yesterday. Found a three other minor issues:

  MM> You didn't tell us your OS (but we now know it's Windows), nor version
  MM> of Emacs (or even Xemacs ??).

I didn't although emacs did:-)

User-Agent: Gnus/5.11 (Gnus v5.11) Emacs/22.0.990 (windows-nt)

I'm using Emacs from Emacs W32, on XP, latest R. 


  PL> Crashes for me with search-failed. Would

  PL> (file-name-directory (locate-library "ess-site.el"))

  PL> than the command history?

  MM> Good idea !  [if it works -- which is not easy to test for me,
  MM>  because for me, the first clause
  MM> 	 (and (boundp 'load-file-name) load-file-name)
  MM>  always triggers i.e. the find-load-file-directory function never needs
  MM>  to be called
  MM> ]

  MM> Your proposal should be amended, however, to

  MM> 	(file-name-directory (locate-library "ess-site"))

  MM> since it can be *.el or *.elc {byte-compiled version}

  MM> Looking at the above (defun find-load-file-directory ....)  I must say
  MM> that it looks very kludgey to me also since it relies on an explicit
  MM> (load-file ...) call which is not at all the only way ess-site may have
  MM> been loaded.

Ah, I think I have just worked out what was going wrong. The load-file-name
form should be returning true for me also. I am guessing here, but I guess
that I did eval-buffer while in ess-site.el. 


  --> I'm trying your suggestion
  MM>     and send you (privately) a changed ess-site.el that you can try, ok?



  PL> Second:

  PL> Had some weird issues with plot windows hanging -- at first I thought
  PL> this was bad, but they seem to be recovering now, so I guess this is not
  PL> a problem.

  MM> Once we know more about your setup (OS version, Emacs version, R version
  MM> ...) people may have hints for you.

Given above. 


  PL> Third:

  PL> Reading through the R tutorial, I have realised that I know almost know
  PL> statistics. Suggestions welcome....

  MM> The book by Peter Dalgaard, or another one from
  MM>    http://www.r-project.org/doc/bib/R-books.html

Thanks! Will do. 

Phil