Bugs

From Phaserwiki
Revision as of 22:37, 1 July 2009 by WikiSysop (talk | contribs) (Known Bugs Phaser-1.2: add icon)

We apologise for the bugs.

If you come across a bug, check the known bug list below and if the problem is new please report it to us.

Bug.pngPhaser-1.3.1

  • Memory usage can go out of control under OSF1 (Tru64 Unix)
    This problem is specific to OSF1, and we believe it arises from a bug in the handling of the standard map library on that platform.
  • Core dump if maximum score is negative (specific to IRIX executable)
    Fixed for upcoming patch release. Work around by (e.g.) increasing RMS to avoid negative scores, or by running under a different operating system..
  • SGALT TEST doesn't work if only one test space group specified
    Fixed for upcoming patch release. Work around by using SPACEGROUP command instead..
  • Using DQ option of NMAPDB command leads to infinite loop
    Fixed for upcoming patch release. Work around by using MAXRMS option instead.
  • If COMPosition is specified by SEQuence, the last line of the sequence file is used twice
    Fixed for upcoming patch release. Work around by adding comment line (line starting with ">") at end of sequence file.
  • COMPosition ENSEmble FRACtional keyword doesn't work
    Work around by using the other methods to specify the composition of the asymmetric unit.
  • ROTATE AROUND option of MR_BRF has disappeared from the ccp4i GUI
    Fixed for upcoming patch release. Work around by entering correct command using the "Run&View Com File" option of ccp4i.

Bug.pngPhaser-1.2

  • .../phaser/src/Ensemble.cc(663): PROGRAM_ASSERT(storedV == mapV[b]) failure.
    This is a numerical instability that has been fixed for the upcoming 1.3 release. If you have the installer release, you can replace line 663 of Ensemble.cc with
    PROGRAM_ASSERT(storedV < mapV[b]+tol && storedV > mapV[b]-tol);
    then recompile.
    Otherwise you can try changing the resolution ever so slightly (e.g. from 2.5Å to 2.51Å) as this might prevent the bug from showing up.
  • Molecular replacement using electron density as a model does not work.
    Bugs were introduced, breaking this feature prior to the release of version 1.2. This has now been fixed for the upcoming release. If you need to use this feature now, please contact us to obtain a pre-release executable.
  • Default fast rotation search returns no orientations.
    The default search should return at least one orientation above 75% of the maximum. If no orientations are returned, you have encountered a numerical stability problem with the fast rotation function that shows up on some architectures (e.g. MacOSX, Irix) when the molecule is large compared to the resolution of the data. This has been fixed for the upcoming 1.3 release. In the meantime, you can work around this problem by: 1) running under Linux, where the problem does not show up; 2) reducing the resolution, as the default is usually higher than needed for large molecules; 3) running with VERBOSE ON to find out how large the sphere is for the Patterson decomposition for the fast rotation, then running again with the CLMN SPHERE set to a smaller number, say 10-20% smaller than the default.