Hello,
I was trying to get the new version, and I had this error:
babiuc$ ./GetComponents -p -a https://svn.einsteintoolkit.org/manifest/trunk/einsteintoolkit.th Can't locate object method "new" via package "Thread::Semaphore" at ./GetComponents line 600.
Apropos of Mac OS, is there a still maintained wiki tutorial on how to get the Toolkit working? I could not find the path to it.
Maria
Hi,
On Tue, May 15, 2012 at 02:38:51PM +0000, Babiuc-Hamilton, Maria wrote:
babiuc$ ./GetComponents -p -a https://svn.einsteintoolkit.org/manifest/trunk/einsteintoolkit.th Can't locate object method "new" via package "Thread::Semaphore" at ./GetComponents line 600.
I am not a Mac user - anyone? Btw: which MacOS and perl versions do you use?
Apropos of Mac OS, is there a still maintained wiki tutorial on how to get the Toolkit working? I could not find the path to it.
I don't know who started it, probably Ian Hinder. However, as I learned today it apparently not only matters which version of Mac OS you use, but also how (which) additional non-Mac libraries you installed (fink vs. macports vs self-compiled).
Frank
On 16 May 2012, at 04:20, Frank Loeffler wrote:
Hi,
On Tue, May 15, 2012 at 02:38:51PM +0000, Babiuc-Hamilton, Maria wrote:
babiuc$ ./GetComponents -p -a https://svn.einsteintoolkit.org/manifest/trunk/einsteintoolkit.th Can't locate object method "new" via package "Thread::Semaphore" at ./GetComponents line 600.
I am not a Mac user - anyone? Btw: which MacOS and perl versions do you use?
Apropos of Mac OS, is there a still maintained wiki tutorial on how to get the Toolkit working? I could not find the path to it.
I don't know who started it, probably Ian Hinder. However, as I learned today it apparently not only matters which version of Mac OS you use, but also how (which) additional non-Mac libraries you installed (fink vs. macports vs self-compiled).
There are 4 tutorials listed under Documentation on the main page of the wiki (https://docs.einsteintoolkit.org/et-docs/Main_Page). However, I am not sure how maintained these are. Also, it was decided long ago that we would not actively support people running on their own machines with tutorials initially, as the number of different configurations was too hard to predict. Instead, the tutorial for new users relies on you running on a "known" machine: Queen Bee.
I think it is time to revise this strategy, and investigate if it is possible to provide a tutorial which is general enough to run on a wider range of machines. It would be good if it was not specific to Queen Bee, and if it could run on someone's local machine. We could give a list of required packaged for several linux versions and Mac OS for laptops/workstations.
These tutorials should also be tested and updated for each release.
Maybe we should base the tutorials on a VM?
Cheers, Steve
On 05/16/2012 02:27 AM, Ian Hinder wrote:
On 16 May 2012, at 04:20, Frank Loeffler wrote:
Hi,
On Tue, May 15, 2012 at 02:38:51PM +0000, Babiuc-Hamilton, Maria wrote:
babiuc$ ./GetComponents -p -a https://svn.einsteintoolkit.org/manifest/trunk/einsteintoolkit.th Can't locate object method "new" via package "Thread::Semaphore" at ./GetComponents line 600.
I am not a Mac user - anyone? Btw: which MacOS and perl versions do you use?
Apropos of Mac OS, is there a still maintained wiki tutorial on how to get the Toolkit working? I could not find the path to it.
I don't know who started it, probably Ian Hinder. However, as I learned today it apparently not only matters which version of Mac OS you use, but also how (which) additional non-Mac libraries you installed (fink vs. macports vs self-compiled).
There are 4 tutorials listed under Documentation on the main page of the wiki (https://docs.einsteintoolkit.org/et-docs/Main_Page). However, I am not sure how maintained these are. Also, it was decided long ago that we would not actively support people running on their own machines with tutorials initially, as the number of different configurations was too hard to predict. Instead, the tutorial for new users relies on you running on a "known" machine: Queen Bee.
I think it is time to revise this strategy, and investigate if it is possible to provide a tutorial which is general enough to run on a wider range of machines. It would be good if it was not specific to Queen Bee, and if it could run on someone's local machine. We could give a list of required packaged for several linux versions and Mac OS for laptops/workstations.
These tutorials should also be tested and updated for each release.
On Wed, May 16, 2012 at 09:11:18AM -0500, Steven R. Brandt wrote:
Maybe we should base the tutorials on a VM?
Yes, I think this is a good idea. Regardless of this I also think that it would be nice to find volunteers that agree to keep some description of how to get it working on a particular, common installation (that they themselves use) up to date. I could imagine that MacOS should be common enough, along with something like the some Fedora or Debian version.
Frank
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 16 May 2012, at 16:15, Frank Loeffler wrote:
On Wed, May 16, 2012 at 09:11:18AM -0500, Steven R. Brandt wrote:
Maybe we should base the tutorials on a VM?
Yes, I think this is a good idea. Regardless of this I also think that it would be nice to find volunteers that agree to keep some description of how to get it working on a particular, common installation (that they themselves use) up to date. I could imagine that MacOS should be common enough, along with something like the some Fedora or Debian version.
A VM is good for workshops and hands-on sessions to teach people how to use the ET. But we also need to provide practical information for how to use the ET on machines that people actually want to use. The main tutorial should in principle be independent of the place it is run.
So we would have a set of "setting up on Mac OS", "setting up on Fedora", "setting up on the ET VM", "setting up on Queen Bee" documents, and then a tutorial which can be followed on any of these systems. We may need a couple of "special case" discriminations in the general tutorial, but hopefully these will be minimal.
Keeping the tutorials up-to-date should be part of the release process, and I think should be considered quite important.
- -- Ian Hinder http://numrel.aei.mpg.de/people/hinder
On Wed, May 16, 2012 at 05:05:54PM +0200, Ian Hinder wrote:
A VM is good for workshops and hands-on sessions to teach people how to use the ET. But we also need to provide practical information for how to use the ET on machines that people actually want to use. The main tutorial should in principle be independent of the place it is run.
In practice this is not true. A lot depends on system versions or compilers being installed, or not.
We may need a couple of "special case" discriminations in the general tutorial, but hopefully these will be minimal.
Keeping the tutorials up-to-date should be part of the release process, and I think should be considered quite important.
I agree with both.
Frank
On 16 May 2012, at 17:26, Frank Loeffler wrote:
On Wed, May 16, 2012 at 05:05:54PM +0200, Ian Hinder wrote:
A VM is good for workshops and hands-on sessions to teach people how to use the ET. But we also need to provide practical information for how to use the ET on machines that people actually want to use. The main tutorial should in principle be independent of the place it is run.
In practice this is not true. A lot depends on system versions or compilers being installed, or not.
Those things should be hidden by SimFactory. We would provide instructions for a "recommended" set of system packages and compilers for each OS, and we would provide optionlists for these.
We may need a couple of "special case" discriminations in the general tutorial, but hopefully these will be minimal.
Keeping the tutorials up-to-date should be part of the release process, and I think should be considered quite important.
I agree with both.
Frank
On Wed, May 16, 2012 at 05:38:30PM +0200, Ian Hinder wrote:
Those things should be hidden by SimFactory. We would provide instructions for a "recommended" set of system packages and compilers for each OS, and we would provide optionlists for these.
Exactly. Simfactory needs to be configured for each new machine, and this set of packages is what is we currently do not know/distribute.
So, coming back to Mac OS what we would need is a set of packages and instructions how to install them, and an simfactory option list could then assume they are installed.
Frank
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 16 May 2012, at 17:46, Frank Loeffler wrote:
On Wed, May 16, 2012 at 05:38:30PM +0200, Ian Hinder wrote:
Those things should be hidden by SimFactory. We would provide instructions for a "recommended" set of system packages and compilers for each OS, and we would provide optionlists for these.
Exactly. Simfactory needs to be configured for each new machine, and this set of packages is what is we currently do not know/distribute.
So, coming back to Mac OS what we would need is a set of packages and instructions how to install them, and an simfactory option list could then assume they are installed.
Sorry, I misunderstood you. I thought you were saying that it was not feasible to provide these instructions. We have actually put instructions in the Mac OS optionlists for which packages you need. I recommend the optionlist https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/osx-l... which has in the comments
# osx-lion-macports-gcc # # Assumes the following are installed by MacPorts: # szip, gcc46, fftw, fftw-3, gsl, openssl, hdf5-18 +gcc46 -universal, zlib, openmpi +gcc46
We also have an optionlist for debian (https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/debia...) which contains these comments:
# Configuration for a Debian installation, assuming the following list of # packages is installed: # # perl gfortran g++ libmpich2-dev # # In addition, installing the following list of packaed will prevent Cactus # from compiling it's own versions of these libraries: # # libfftw3-dev libgsl0-dev libatlas-base-dev libjpeg-dev libssl-dev libhdf5-serial-dev # # Tools like GetComponents and Simfactory like to have the following list # installed too # # python subversion git mercurial
and a Fedora optionlist (https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/fedor...) with these comments:
# Fedora # # Yum install command line needed to make this work: # # yum -y install mpich2 mpich2-devel \ # gsl gsl-devel hdf5 hdf5-devel \ # gcc gcc-c++ gcc-gfortran patch \ # java-1.6.0-openjdk subversion \ # git openssl-devel
I would say that we are already in pretty good shape! If we cannot find volunteers to test these optionlists and commands on the most recent release of each of these systems at the time of the ET releases, we could also do this using virtual machines (for linux, anyway).
- -- Ian Hinder http://numrel.aei.mpg.de/people/hinder
On Wed, May 16, 2012 at 05:52:51PM +0200, Ian Hinder wrote:
# Assumes the following are installed by MacPorts: # szip, gcc46, fftw, fftw-3, gsl, openssl, hdf5-18 +gcc46 -universal, zlib, openmpi +gcc46
It should probably also mention that you should install compilers, e.g. using XCode (or whatever should be used?).
We also have an optionlist for debian (https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/debia...)
Yes - I compiled this, and this is what is also used for the virtual machine.
and a Fedora optionlist (https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/fedor...) with these comments:
Who is willing to maintain this?
Frank
On Wed, May 16, 2012 at 5:16 PM, Frank Loeffler knarf@cct.lsu.edu wrote:
On Wed, May 16, 2012 at 05:52:51PM +0200, Ian Hinder wrote:
# Assumes the following are installed by MacPorts: # szip, gcc46, fftw, fftw-3, gsl, openssl, hdf5-18 +gcc46 -universal,
zlib, openmpi +gcc46
It should probably also mention that you should install compilers, e.g. using XCode (or whatever should be used?).
The necessary compilers are included in the MacPorts package list.
MacPorts does still require that XCode is installed (to compile the compilers), but that is covered by the MacPorts installation.
I finished installing on betazoid here at CCT. Here's a complete log. I used Barry's config file, but also put in many additional notes:
% cat install_notes.txt
- OS X 10.7.4
I. Xcode: ------------ -Xcode: cf. http://guide.macports.org/chunked/installing.xcode.html - XCode from App Store (free). Current version is 4.3.2, but apparently 3.2.6 was already installed. MacPorts (next) was detecting the 3.2.6 version and refused to run. Had to go into /Applications and run % open -a Xcode which brought up 4.3.2, with some installation dialog boxes. After that, since /Developer was used for previous versions but not anymore, did a % sudo mv /Developer /Developer.O Also had to run % sudo xcode-select -switch /Applications/Xcode.app/Contents/Developer Then add the command line tools: Run Xcode, go to Preferences > Downloads
II. MacPorts: ----------------- - MacPorts binary package from MacPorts website
( - Optinal: Pallet: (also a good test to see if MacPorts is set up properly) % sudo port install pallet ...and say yes to install Java runtime. Pallet then appears in the /Applications directory, or can be run from command line via % open -a pallet (This Mac's filesystem is case-insensitive btw) Pallet doesn't actually work well on Lion for installing packages, but can be useful to get a quick list of packages. Installation of other packages should still be done using "port" command. )
- add /opt/local/bin to PATH.
III. MacPorts Packages ---------------------------- - Following Barry Wardell's notes, adding the revision control packages: % sudo port install szip gcc46 fftw fftw-3 gsl openssl hdf5-18 +gcc46 -universal zlib openmpi +gcc46 mercurial git-core subversion cvs
IV. Einstein Toolkit (ETK) ----------------------------------- - Downloaded ETK as per Wiki Download page. Used current development version - Built via simfactory (because I wanted to learn how to do that): Created a simple machine file: ~Cactus/simfactory/mdb/machines/betazoid.ini based on Peter's file for his laptop - In betazoid.ini, referenced Barry's osx-lion-macports-gcc.cfg file Then built: % cd ~/Cactus; simfactory/bin/sim build etk --thornlist=thornlists/einsteintoolkit.th
- Ran qc0-mclachlan: ( Optional: Use OpenMP: % setenv OMP_NUM_THREADS 4 ) % cd ~/Cactus/par; ../exe/cactus_etk qc0-mclachlan.par I changed cctk_final_time to 1.0 instead of 100.0 so the thing would finish quickly... Should have set it to 0.25!
V. Visit & CarpetHDF5 ------------------------------- - Installed 64-bit Mac binary package of Visit 2.4.2 from Visit website. - Downloaded CarpetHDF5 as per website % cd ~; svn co https://svn.cactuscode.org/VizTools/CarpetHDF5/trunk/ CarpetHDF5 % cd ~/CarpetHDF5 % setenv PATH "/Applications/VisIt.app/Contents/Resources/bin/ $PATH" // I use tcsh :-)
- Need to get Cmake: % sudo port install cmake
- And since the binary of Visit expects various libraries in /usr/lib/, we do % sudo ln -s /opt/local/lib/libmpi* /opt/local/lib/libopen-rte.* /opt/local/lib/libopen-pal.* /usr/lib ..Feel free to compile Visit from source if you want to skip this step...
- Then run the install script % ./install Note that the window appears *under* the terminal window you're using! Click the CMake tab, and set /Applications/VisIt.app/Contents/Resources/2.4.2/darwin-x86_64/ CXXFLAGS = -I/Applications/VisIt.app/Contents/Resources/2.4.2/darwin-x86_64/include -I/opt/local/include LDFLAGS = -L/Applications/VisIt.app/Contents/Resources/2.4.2/darwin-x86_64/lib -L/opt/local/lib
- Saved the file, quit xmledit and the plugin builds!
- Run visit with the -noconfig option. % cd ~/Cactus/par/qc0-mclachlan/; visit -noconfig
- Choose, e.g dambase::lapse.h5 It Works! :-)
Hello,
For MacOSX, Macports is the easier way to get all the software needed. My perl version is 5.12 and python 27. I was able to checkout a new version, and update an old one when I did not use the -p The old instructions for Mac OSX were working great, if we can update then would be fantastic. I'm using gcc45 and the old configuration still works. Haven't tried with gcc48. Did anyone? Scott came up already with new instructions for Visit, still to be confirmed. Barry Wardell has made possible the installation of pygraph, which works.
Maria ________________________________________ From: users-bounces@einsteintoolkit.org [users-bounces@einsteintoolkit.org] on behalf of Frank Loeffler [knarf@cct.lsu.edu] Sent: Wednesday, May 16, 2012 11:46 AM To: Ian Hinder Cc: users@einsteintoolkit.org Subject: Re: [Users] Einstein Toolkit on Mac OS
On Wed, May 16, 2012 at 05:38:30PM +0200, Ian Hinder wrote:
Those things should be hidden by SimFactory. We would provide instructions for a "recommended" set of system packages and compilers for each OS, and we would provide optionlists for these.
Exactly. Simfactory needs to be configured for each new machine, and this set of packages is what is we currently do not know/distribute.
So, coming back to Mac OS what we would need is a set of packages and instructions how to install them, and an simfactory option list could then assume they are installed.
Frank
Frank and I are currently working to install on a somewhat "fresh" Mac at CCT. I'm making a list of all the packages, etc. that I install.
-Scott
On 5/16/12 11:14 AM, "Babiuc-Hamilton, Maria" babiuc@marshall.edu wrote:
Hello,
For MacOSX, Macports is the easier way to get all the software needed. My perl version is 5.12 and python 27. I was able to checkout a new version, and update an old one when I did not use the -p The old instructions for Mac OSX were working great, if we can update then would be fantastic. I'm using gcc45 and the old configuration still works. Haven't tried with gcc48. Did anyone? Scott came up already with new instructions for Visit, still to be confirmed. Barry Wardell has made possible the installation of pygraph, which works.
Maria ________________________________________ From: users-bounces@einsteintoolkit.org [users-bounces@einsteintoolkit.org] on behalf of Frank Loeffler [knarf@cct.lsu.edu] Sent: Wednesday, May 16, 2012 11:46 AM To: Ian Hinder Cc: users@einsteintoolkit.org Subject: Re: [Users] Einstein Toolkit on Mac OS
On Wed, May 16, 2012 at 05:38:30PM +0200, Ian Hinder wrote:
Those things should be hidden by SimFactory. We would provide instructions for a "recommended" set of system packages and compilers for each OS, and we would provide optionlists for these.
Exactly. Simfactory needs to be configured for each new machine, and this set of packages is what is we currently do not know/distribute.
So, coming back to Mac OS what we would need is a set of packages and instructions how to install them, and an simfactory option list could then assume they are installed.
Frank
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On Wed, May 16, 2012 at 5:16 PM, Scott Hawley scott.hawley@belmont.eduwrote:
Frank and I are currently working to install on a somewhat "fresh" Mac at CCT. I'm making a list of all the packages, etc. that I install.
I did the same thing a couple of months ago when I created the osx-lion-macports-gcc.cfg SimFactory optionlist, so I can't imagine the instructions being very different from what is mentioned in the comments at the top of that file. If there are things which have changed, it would be good to also update the comment in the optionlist.
The only thing I didn't use was GetComponents as I used a different method for getting my Cactus tree.
Barry
users@lists.einsteintoolkit.org
-
Babiuc-Hamilton, Maria -
Barry Wardell -
Frank Loeffler -
Ian Hinder -
Scott Hawley -
Steven R. Brandt