2.0RC3

Release Date

2007-10-19

Summary

This is the third release candidate for 2.0. All major bugs we are aware of have been fixed. A number of very critical fixes have been made in this RC version.

Important Note: PLEASE read the release notes in the CHANGELOG.

Description

NOTES:

• PLEASE LOOK AT THE INSTALLATION TIPS : appropriate Wiki page
• If you run into any problems, please complain or use our bug tracker.
• The windows version is 32 bit. But it can also work on 64 bit windows platform. It has been tested for WindowsXp 32/64, WindowsVista 32/64, and Windows 7 32/64. There may still be difficulties with the Windows version, if you do encounter problems, please let us know. Please refer to more details about the installation on wiki
• Documentation is growing, and provided through the Wiki.
• These distributions also include SPARX
• While the linux binaries could potentially work with Linux versions as old as Fedora Core 3, a more recent release is HIGHLY recommended.
• The GCC4 binaries provide a significant performance boost. It needs glibc version at least 2.5. You can check your system's glibc version by using command '/sbin/ldconfig -V'.
• When upgrading from an older version, BEFORE untaring the new version, make sure you run 'e2bdb.py -c' to flush and clean up the cache. This will also make sure you don't have any EMAN2 jobs running while you upgrade. If you fail to do this, you may need to 'rm -rf /tmp/eman2db-*' before you will be able to start new EMAN2 jobs, and there is a (small) possibility of some file corruption.

Version changes

• Major changes since RC2
  • MAJOR FIX - the 3-D reconstruction algorithm could produce streaked or stretched looking reconstructions when keep<1 in e2make3d. This problem has existed for ~8 months and has now been fixed. If you observed a funny looking reconstruction using a version from the last 8 months or so, please try re-running the refinement with this new version !!!
  • Anisotropic Wiener filter in 3-D reconstruction now supported
  • Easy to use multi-core parallelism without use of client/server model
  • Fix for memory leak when running in parallel on a cluster
  • Extended structure factor determination closer to origin. Recompute structure factors with the new version.
• KNOWN BUGS
  • On some Win XP/7 machines we have found that switching between images in boxer causes Swarm to stop functioning. Only completely exiting the workflow and restarting seems to permit boxing a new image. If you experience this, please report it