[Ifeffit] ifeffit and Mac OSX

Bruce Ravel ravel at phys.washington.edu
Thu Oct 10 10:49:50 CDT 2002


 MN> Is the Fink / X-Windows / PGPLOT installation procedure too big
 MN> of a hassle?  I know that can work, and that athena will work
 MN> too.  I also believe that the instructions are simpler for 10.2
 MN> than 10.1 (especially with respect to getting perl/Tk to work),
 MN> but haven't tried this myself.

 MN> I'm copying this message to the ifeffit mailing list, hoping that
 MN> someone else may have better answers to these questions.

 MN> How necessary is a complete binary installer for Mac OSX?

I am really glad this issue came up (even if it took me a few days to
return home from the sychrotron, unwind, then begin to hack at my
piles of email).

While I was at NSLS over the last weekend, I made a stab at following
the instructions on a iBook running 10.1.  Things did not go well for
me.  I think the real problem was that I was behind a firewall and
could not figure out how to coerce fink into doing its thing through a
proxy.  (Yes, I followed the instructions in the FAQ -- still didn't
work.  Then I had to get back to my experiment.)  I was able to
download and install fink without any trouble, but then, as I said,
had trouble configuring it for the place I was at.

There was the additional problem that owner of the computer was
bewildered by the entire concept.  He basically understood the notion
that Athena depended on other software that had to be installed first,
and he basically understood the concept of fink being a download
manager, but I doubt that he would be able to find his way from the
"download fink" step to the "run athena" step.  That's a problem.

There is a solution.

Fink is just debian's apt-get tidied up for the needs of OSX and the
Mac hardware.  apt-get is a wonderfully versatile utility.  While it
is usually run from a modal UI like dselect, it also has a full
command-line interface and can be incorporated into a shell script or
any programming language with a system call.

What's more you can tell apt-get to go fetch its packages from
anywhere you want.  Typically there are some default repositories from
which you normally fetch stuff (that is certainly the case with fink's
dselect -- it goes to a place in the sourceforge universe where
packages for OSX live).  However, you can tell apt-get to explicitly
fetch packages from any place on the web.  That's the key.

Here is my proposal:

  1.  One figures out the minimal set of packages (i.e. the x-windows
      stuff and so on) that are needed on the system before starting
      with ifeffit.  Presumably there are binary packages for these.
  2.  Ifeffit, PGPLOT, and the GUIs (mine, Matt's, Sam's) should be
      compiled for 10.1, 10.2 and any other versions commonly lurking
      out there.  These should be packaged into binary .debs.
  3.  The Ifeffit etc binary .debs should be housed on a known
      location on the web, such as the Ifeffit web site.
  4.  A script gets written that, assuming fink has been installed,
      uses apt-get directly to fetch the necessary files from the
      standard places, then fetches Ifeffit, etc from its repository.

This reduces the installation instructions to three steps:

  1.  connect to the web
  2.  download and install fink
  3.  run (double click on?) a script that runs apt-get to install a
      pre-defined set of binary .debs

Except for the firewall issue, that should be simple enough even for
(flamebait alert!)  Macintosh users.

Someone other than me is going to have to act on this.  I simply do
not have a Macintosh on my desk and, at the moment, lack the money
(either personally or professionally) to buy one.  There are several
Mac gurus reading this email who have already done much of what is
required.

If this issue is not addressed, I will be forced to remove language
from my web page stating that the software is supported on OSX.  That
would make me sad.

B

P.S. I suppose that binary .debs are not needed.  apt-get is certainly
capable of dealing with source code and doing configure and make.
However, the same concept would apply.  Source .debs would have to be
made and collected into a repository that the instalation script knows
how to look at.


-- 
 Bruce Ravel  ----------------------------------- ravel at phys.washington.edu
 Code 6134, Building 3, Room 222
 Naval Research Laboratory                          phone: (1) 202 767 5947
 Washington DC 20375, USA                             fax: (1) 202 767 1697

 NRL Synchrotron Radiation Consortium (NRL-SRC)
 Beamlines X11a, X11b, X23b, X24c, U4b
 National Synchrotron Light Source
 Brookhaven National Laboratory, Upton, NY 11973

 My homepage:    http://feff.phys.washington.edu/~ravel 
 EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/



More information about the Ifeffit mailing list