Known bugs

• The command SET GRAPHICS BOUNDINGBOX will cause a memory-leak bug on some SGI graphics systems, like the Indigo XS.

• If a crosspeak file is owned by someone else than the user running ANSIG, then the program will most likely crash as ANSIG tries to write something into the crosspeak file. The error message is:

  Error: reading from crosspeaks file
  no crosspeaks retained
  close: Bad file number
  apparent state: unit 1 named crosspeakfilename
  lately writing direct formatted external IO
  *** Execution Terminated (9) ***
  

  This problem appears to be due to some bug in IRIX 5.3.

• In some cases an infinite loop is created when crosspeaks in different 3D and 4D spectra are connected to each other. Unfortunately, it is necessary to kill the ANSIG process (ctrl-C or the UNIX kill command) and check the crosspeak database with the help of the command CROSSPEAKS REPAIR.

• When the view dimensions of a spectrum are set to the reverse of the default, then setting a crosspeak in this spectrum to be corresponding to a crosspeak in another spectrum will most likely not create the connection that was intended. Strange transfer of assignments may then also occur.

• When changing the view of spectrum in a window, then the loaded crosspeaks for that spectrum are not loaded to that window again properly, unless those crosspeaks already exist in another window with the same view.

• The implementation of the 1D trace graphics window has not been finished. The ruler position is determined by both 2D rulers, resulting in confusion. Currently the 1D trace window misbehaves with 3D and 4D spectra; sometimes it refuses to extract the trace. The ruler should be pickable. There should be a set of AL intrinsic procedures for the 1D window.

• When moving crosspeaks interactively, folded crosspeaks are not handled correctly. Aliased crosspeaks are handled correctly.

• It is currently not possible to control which spectrum dimension gets mapped to which non-plane scroll bar in a window, in those cases when there is potentially a choice. For instance, with double-labelled protein, it is possible to end up with spectra where the first non-plane scroll bar controls a 15N dimension in one spectrum, while it controls a 13C dimension in another spectrum, which of course does not make sense.

• The automatic peak-picking for 3D and 4D spectra has not been implemented yet. The current automatic 2D peak-picker is probably not optimal. How should rectangular and diagonal masks work for 3D and 4D?

• The command CROSSPEAKS STORE does not save the connections between the crosspeaks in the same spectrum; it should.

• The crosspeak integration method implemented is too coarse; something more sophisticated is needed.

• 3D homonuclear NOESY-NOSY or NOESY-TOCSY spectra have not been tested with this version of ANSIG, so although they should work, there may be unknown problems with them.