Snow Leopard is Here

by Jessica Lu on August 28, 2009

The latest and greatest version of Mac OS X, dubbed Snow Leopard, is here. This is likely to be  a good release for astronomers as the OS has been trimmed down to boost performance overall and to give you back about 7 Gbytes of disk space. However, only Intel Macs are supported. Any early adopters out there??? If so, post your experiences with installation, favorite new features, and astronomer software on Snow Leopard in the comments below.

{ 15 comments… read them below or add one }

1 Sammy August 29, 2009 at 8:33 am

In general, Snow Leopard feels a little snappier than Leopard. Installation was simple; DVD in and few clicks. However, if you wish to do a clean install, be ready to use Disk Utility.
Expose now shows the name of each window below, which is a nice addition, especially if one has multiple windows and terminals open. I have ran into few incompatible programs, namely istat pro (the one that lives in task bar) and a mail plug-in.
I haven’t had change to use IRAF or PyRAF yet, so I cannot comment on those. Not sure what the default Python is now, as I have highly customized setup, but it would be great to have a 64-bit version. Anyone care to comment Python, PyRAF issues?

Reply

2 Don August 29, 2009 at 11:01 pm

I’ve installed Snow Leopard on my Mac Pro and my MacBook Pro. The IRAF ‘cl’ crashes on both. After checking the console logs, it appears that the location of the iraf.h file is lost? The quickest fix I’ve found so far is to login as the IRAF administrator and re-run the install.

I also ran the mkiraf command and then deleted the new login.cl file and replaced it with my original login.cl file. (I didn’t back up my original, IRAF just renames it with the extension .OLD). I’m not certain if running the mkiraf command again contributes to solving the issue; however, the cl parameter file also appeared to be missing; so, running it couldn’t hurt anything. All is running flawlessly again.

Aside from that minor issue, Snow Leopard is definitely living up to the claims. There is a marked increase in performance on all three computers that I’ve installed it on so far.

Reply

3 Eli August 30, 2009 at 2:44 am

I have not tried out PyRAF on Snow Leopard (SL), but have been actively investigating how Python behaves on the new system. It has Python 2.6.1 as its default setting. Depending on CPU your computer has, 32-bit or 64-bit, it will default to 64-bit if it can. You can verify this by typing “file `which python`” in the terminal.

A surprising feature for Python was easy_install. This is a great tool for installing and upgrading your python modules.

I also tried to build Python from source to install more complicated modules. So far no success.

Basic X11 dependent programs worked.

Reply

4 Sammy August 30, 2009 at 9:52 am

Few notes on Python modules. At least on my machine, NumPy SVN head fails to compile. The problem seems to be related to a math library being missing (perhaps a new location). Thus, you might have to edit the MATHLIB env variable before it compiles. I haven’t investigated in detail yet, so it might compile OK with default setup.
matplolib 1.0.snv does not compile without problems either. There are warnings about architecture problems. I also noted following line from the log: “Compiling with an SDK that doesn’t seem to exist”. Perhaps some compiler and Xcode locations have changed and they are not found. I also noted many references to OSX 10.3 while trying to compile matplotlib from the SVN.

Reply

5 Eli August 30, 2009 at 3:13 pm

Sammy, NumPy (SVN) worked without issues for me. I installed it and did a numpy.test(‘1′,’10’) with no errors. I did have errors with SciPy though. If I compile it 32 bit it installs, but the binary type does match (32 bit to 64 bit Python). 64 bit does not work though. It looks like a potential issue with my Fortran compiler. I have not tried matplotlib yet.

Are you using Snow Leopard’s (SL) system Python or did you install a Framework Python setup?

Reply

6 Stuart Littlefair September 2, 2009 at 8:38 am

Probably too obvious to mention, but I have had all sorts of problems compiling code under Snow Leopard until I realised the default was to create 64-bit object files, which wouldn’t link against my (many) 32-bit libraries I have which were compiled under Leopard. Adding the -arch i386 flag fixed my woes.

No problems so far with the Scisoft OS X versions of python or pyraf..

Reply

7 Sammy September 3, 2009 at 10:36 am

Eli, I have multiple Python setups on my laptop. So far I have tried with the Framework setup, where I have 3.1.1 (only for pure Python) and 2.6 (which used to be my work horse before SL). I haven’t had time to try with the SL system Python, but I assume that things work better with that one. However, there seems to be some confusion with the type mismatch as you noted.
I also have Enthought Python installed, but I have tried to separate it from the system Python. Enthought seems to work fine even after the SL upgrade.

Reply

8 Eli September 10, 2009 at 7:07 am

Sammy, one more update. I got NumPy, SciPy, and Matplotlib all working on Snow Leopard now. There are some issues with specific packages in SciPy, but overall it works very well thanks to the following blog post: http://blog.hyperjeff.net/?p=160

Reply

9 Sammy September 10, 2009 at 8:55 am

Eli, I noticed the same excellent post few days ago. However, I still have issues with gcc while compiling NumPy. I should check my version of gcc and other compilers and libraries to see where they differ with the default SL installation. It also seems that when compiling 64-bit Python Modules my setup finds a wrong architecture libpython2.6.dylib. I guess this is the pain of having four separate Python installations…

Reply

10 Sammy September 10, 2009 at 9:58 am

Sorry to spam again, but seemed that my problems were solved by getting rid off the 2.6.2 Python I had installed. This version seemed to somehow confuse the Framework default 2.6.1 that SL has and cause some architecture mismatches.

Reply

11 Nate September 15, 2009 at 5:14 pm

I waited to buy my new desktop until SL was shipping with new Macs. Today I installed IDL 6.3, and find that it crashes if I try to display an image using tv, tvscl, display, etc. I had similar problems crop up under OS X 10.4 occasionally (after automatic MacUpdates), but installing the latest update of X11 always fixed it. Now X11 (XQuartz 2.3.4) is native to OS X, and there is no SL-supported update as of yet (XQuartz 2.4.0 is for regular Leopard). So, it appears I have an installed IDL that cannot display images. Which is mainly what I work with. Any ideas?

Reply

12 Jonathan September 22, 2009 at 8:39 pm

All right, last week I just purchased my first mac ever — a 15″ MacBook Pro — came with Snow Leopard. (previously used ubuntu). I’ve been trying to get scisoft distro and the enthought distro to work on my computer. I can get some things in scisoft to work, but never pyraf.
I can get Enthought to work perfectly with everything it comes with, until I try to install/reinstall scisoft, which breaks it almost entirely.

So, I’ve settled for, perfect Enthought distro; non-working scisoft, but I wanto this situation to improve.

My guess is to separate the python distributions (maybe three ways, scisoft/enthought/system?) but I have no idea how to do this. I saw a previous comment that mentioned having separate pythons doing something like this, and I’d like to know how to do this with the distrobutions enthought and scisoft.

Any tips? Ideas? Suggestions to radically change my approach?

Reply

13 Jeremy October 5, 2009 at 12:12 am

Nate,
I had the same problem with images in IDL 6.4 on a mac pro that
started a few months ago (presumably as a result of an auto software
update or something). On a TVSCL or similar, IDL crashed with a bus
error. Have just installed SL, and the problem persisted
with IDL 6.4. Installed 7.1 and problem seems to be fixed.
cheers,
Jeremy

Reply

14 Lisa December 2, 2009 at 12:00 pm

I was able to install iraf/pyraf and matplotlib (I had a problem where I needed to set the architecture to 32-bit). I’m having problems with plotting in both, though. In matplotlib, I can successfully import pylab, but am unable to plot. I’m using the MacOSX backend. It will open a blank window labeled Figure 1, but not actually plot anything. If I try to save the plot to a ps file, the file has errors and will not open.

With pyraf, I am having the problem that when I use implot/splot there are no labels on the axes and no titles. The commands in the window work, but I have no labels.

Has anyone else run into any of these problems and been able to fix them?

Reply

15 David April 14, 2010 at 12:18 pm

As a note regarding IDL and X11: as noted above, X11 with SL comes is 2.3.4. X11 2.4 does not have a SL installer (Leopard only) that I know of. X11 2.5 (which, btw, is now called Xquartz when installed and does not overwrite the X11.app) installs on 2.5, but IDL’s mouse-plotting window interaction is broken (e.g., zoom does not work properly). So I went back to the X11 2.3.4 which came with SL.

Reply

Leave a Comment

Previous post:

Next post: