The following bugs were fixed in FETCHAN V6.0.1
===============================================
  
*  When adjusting the length of a detected event in Events List or  
   Latency analysis, it is no longer possible for the adjusted event  
   to wrap around (incorrectly) from the end of one trace to before  
   the beginning of the same trace.  
  
*  When adjusting the trigger threshold in Pulse Analysis, the  
   program no longer crashes when the threshold is within 4 pixels of  
   the maximum or minimum signed integer.  
  
*  Fetchan now correctly displays and plots data files that had  
   been acquired with a negative display gain.  Previously, these  
   files had been displayed and plotted with inverse polarity.  
  
*  When saving the results of an Episode Average to an ATF file,  
   the ATF header lines listing the episodes included in the average  
   now contain no more than 150 episodes per line.  Note:  multiple  
   lines are allowed.  (This was not a Fetchan bug, but had caused  
   problems when other pCLAMP programs tried to read this ATF file.)  
  
*  The demo version of Fetchan now allows saving parameter file  
   changes.  

*  A "Demo" banner is displayed across all plots; even print  
   previews.  
  
  
*  The demo version of Fetchan now does not allow any kind of file  
   export.  Print preview is still allowed.  
  
*  All pCLAMP programs had File/GetInfo enhanced slightly.  

*  The demo version of Fetchan no longer shows a "Demo" banner  
   when performing a Print Preview.  
  
*  Events truncated by the end of analysis region are now included in  
   the events list file.  A new note ("T" for truncate) is used to  
   mark these events.  The EVL version number was increased from 2 to  
   3 due to this change.  Previously, Fetchan had excluded such  
   events so that pSTAT could include all events in its histograms.   
   However, this prevented Popen from being determined correctly.   
   Now, pSTAT will use the "T" flag to exclude truncated events from  
   its histograms.  A new version of READEVNT was also created to  
   read the new EVL files.  
  
  
*  Writing out statistics during Cursor Measure no longer causes  
   the program to crash if the cursors are more than 32,767 ms apart.   
  
*  During Events List or Latency analysis, Fetchan now correctly  
   handles the (very rare) case below.  Previously, it was possible  
   for Fetchan to write out 0 length events to the events list file  
   (such events are impossible, by definition).  This only occurred  
   under the following conditions:  
  
   - The user was accepting all events (Ignore gap duration was 0).  
   - The thresholds were being updated.  
   - An event transition's data point value had exactly the same  
     value as the detection threshold.  
   - When updating the threshold, the threshold moved so that the  
     data point appeared to not have crossed the threshold.  
  
   Under these conditions, Fetchan thought that a transition had  
   occurred twice at the same data point.  This occurred because  
   Fetchan tried to determine if a transition had occurred by  
   examining the transition data point before and after the threshold  
   was updated.  Now, Fetchan only examines the transition point  
   before the threshold is updated.  
  
*  Fetchan no longer allows the following bug to occur in Events  
   List or Latency analysis.  It does so by only allowing one episode  
   per trace to be displayed during these analysis modes.   
   Previously,  Fetchan did not correctly reset the variables used  
   for calculating the mean and standard deviation of an event under  
   the following conditions:  
  
   - More than one episode was being displayed per trace.  
   - For events occurring in the second (or more) episode per  
     trace, the first event in the episode was at the same level as the  
     last event in the previous episode.  
  
   This caused the mean and standard deviation of these events to be  
   incorrectly written out to the events list file.  Assuming that  
   the first and last events in an episode are most likely to be  
   level 0 events, this would affect the level 0 amplitude histogram.   
   As level 0 events generally have amplitudes close to zero, the  
   effect of this bug should be minimal.  
  
   NOTE:  The durations of all events were not affected by this bug.  
  
  
The following bugs were fixed in FETCHAN V6.0.2
===============================================
  
*  Fetchan can now read IPROC files properly.  Previously, it could  
   not read them at all.  
  
*  When in an analysis mode that forces the display window length to  
   be no more than one episode at a time (Episode Average, Segmented  
   Average, Latency), Fetchan now correctly displays the actual  
   window length in the Parameters/DisplayOpts form.  Previously,  
   it would display the length corresponding to the user's requested  
   window length.

*  There have been two options added to the baseline threshold updating
   algorithm, ABF files modules have been updated, and the undo command 
   in Latency/events list has been fixed.

*  Conservative updating.  This is controlled by a menu item located in
   Parameters/special. The menu item is called 'Update strategy'.

   To prevent very short baseline events from biasing the baseline update,
   the user can optionally enable Conservative updating.  Currently, the
   baseline is updated by the following weighted average:

   B_new = B_old(1-F_user)+B_event*f_user

   where B_new is the new value of baseline, B_old is the previous value,
   B_event is the mean of the sampled points contributing to the current
   baseline event and F_user is the fractional weight given to the 
   current baseline event from the user.  In 'conservative updating', this 
   formula would be modified to linearly weight short events as follows:

   F_Cons_user = F_user(N_event/N_cons),  for N_event < N_cons

   F_Cons_user = F_user                   otherwise,

   where N_event is the number of sampled points in the event, N_cons is 
   max (50,100*Tau) (see Note), and Tau is the number of sample points 
   corresponding to the combined filter constant.  The 50 points minimum is 
   to minimize the speed of the baseline updates in case little or no 
   filtering was used, or in case the file header does not contain the 
   correct filter settings.  This correction factor would modify equation 1 
   to be:

        B_new = B_old(1-F_Cons_user)+B_event*F_cons-user.

        Note:   In this test version there are 2 menu items under
                Parameters/special, (POINTS, TAU_Muliplier). These
                correspond to the (50,100*Tau) field.  These items 
                may be changed for tuning purposes.  Also in the event
                analysis screen, there are three new display Items in
                the upper left corner of the screen:
                1. Tau, this is the number of samples ignored because 
                   of filtering effects.
                2. Tau*Mult, this Tau multiplied by the field 'TAU
                   MULTIPLIER'.
                3. Ncons, the user defined number entered into the 
                   field 'POINTS'.


*  To prevent the baseline estimate from becoming obsolete during long
   intervals without events, the user can optionally change the update
   frequency to allow updating during long baseline intervals.  The
   algorithm would then change the definition of N_event to be either the
   number of points in the next baseline event, or n_cons whichever comes
   first.  if N_cons goes by without Fetchan detecting an opening event,
   it will treat that block of N_cons points as being equivalent to an event
   for the puppies of baseline update, and will use equation 1 for the
   calculation (equation 1 and 3 are equivalent in this case).
   The menu item for this is called 'Baseline updating).

*  The last change is to add the new ABF (Axon Binary File) file support
   revision. This adds Macintosh compatibility.

*  The undo command failed to properly restore the search thresholds when 
   used.  This caused all events after the undo to be biased by the event 
   that was undone.

*  During Latency Analysis, if the user rejected the last event in an epoch,
   the next event was labeled rejected also.  Fetchan now handles this situation 
   correctly.  

*  Fetchan now handles the problem of an event ending on the last sample
   of an epoch (or its analysis region).  It will no longer report a
   non-existent event that is 0 time units long

*  When truncated events were undone, Latency analysis would get lost.  All 
   events are now handled by undo correctly.

*  Inside the Latency Analysis window, the event length indicator always 
   displayed 0ms.  This now works correctly.

*  During Latency Analysis, when Ignore was used on the last event before
   reaching the end of the Analysis region, level tracking was not reset
   before the beginning of the next episode.  This feature now works
   correctly.

*  During Latency Analysis, when using ignore past the end of the
   Analysis region, the event level written was wrong.  This feature is
   now working correctly.

*  Set 'Baseline Update Strategy' and 'Update During Baseline' defaults to 
   'wait short events less' and 'Yes', respectively.
   
*  When exporting ASCII data recorded in 'variable-length event' mode 
   with more than one episode per trace, the data times for episodes after 
   the first episode per trace did not match the data times measured on 
   the screen.  This problem has been fixed.


The following bugs were fixed in FETCHAN V6.0.3
===============================================
  
*  Used incorrect version number in releasing 6.0.2.8.  This version
   uses the correct version number.

*  Baseline would get lost on some files that had Baseline tracking
   enabled. These files have extremely long baselines with no events and 
   had thresholds set at the ADC rails.  

*  When writing out the .AV? file, any traces that were excluded before
   the first included trace caused extra 'Episodeavg' lines to be inserted
   into the file.  This caused a problem when using the file as a subtraction
   file.  This problem has now been fixed.

*  Fetchan did not display the amplitude on the scale bar when plotting in
   landscape.  This is now working.

The following bugs were fixed in FETCHAN V6.0.4
===============================================

*  There were no bugs to fix in this release.


