AFNI HISTORY

The levels of importance go from 1 to 5, with meanings:
       1 - users would not care
       2 - of little importance, though some users might care
       3 - fairly important
       4 - a big change or new program
       5 - IMPORTANT: we expect users to know

generated via the command : afni_history -html -reverse -type BUG_FIX 

  ----  log of AFNI updates (most recent first)  ----

18 Sep 2008, RW Cox, Vecwarp, level 1 (MICRO), type 4 (BUG_FIX)
    Only require +orig dataset if user actually NEEDS it

    Program required the +orig version of the -apar dataset, which is needed
    for SureFit work, even if it wasn't actually going to be used.  Not any
    more.


16 Sep 2008, RC Reynolds, 3drefit, level 2 (MINOR), type 4 (BUG_FIX)
    allow attribute editing of NIfTI datasets


10 Sep 2008, RC Reynolds, plug_realtime, level 1 (MICRO), type 4 (BUG_FIX)
    re-added sending of magic_bye string on MP socket close


02 Sep 2008, ZS Saad, 3dSkullStrip, level 2 (MINOR), type 4 (BUG_FIX)
    Starting sphere center was incorrectly initialized


29 Aug 2008, ZS Saad, 3ddelay, level 2 (MINOR), type 4 (BUG_FIX)
    Unitialized pointer in 3ddelay


28 Aug 2008, RW Cox, miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Fixed 'is used uninitialized' errors in several codes.

    Via the new macro ZZME() in 3ddata.h, which zeros out a struct.


27 Aug 2008, DR Glen, 3danisosmooth, level 2 (MINOR), type 4 (BUG_FIX)
    Initialized variable for 3D case


26 Aug 2008, RC Reynolds, 3dAllineate, level 2 (MINOR), type 4 (BUG_FIX)
    initialized ntask in all cases


15 Aug 2008, RC Reynolds, afni, level 2 (MINOR), type 4 (BUG_FIX)
    init graph->grid_spacing, to prevent potential div by 0 via DRIVE


14 Aug 2008, RC Reynolds, 3dBrickStat, level 2 (MINOR), type 4 (BUG_FIX)
    do not automatically print -max along with -var


05 Aug 2008, RC Reynolds, to3d, level 2 (MINOR), type 4 (BUG_FIX)
    re-added the un16 fix from July 1


28 Jul 2008, RC Reynolds, plug_realtime, level 2 (MINOR), type 4 (BUG_FIX)
    alter check for bad socket: use tcp_alivecheck over tcp_writecheck


15 Jul 2008, RW Cox, count, level 1 (MICRO), type 4 (BUG_FIX)
    Make '-suffix' work correctly for last item (per Fred Tam).

    Also, make '-sep' and '-suffix' work as separate items,
    instead of as synonyms for the same thing, which is stoopid.

15 Jul 2008, DR Glen, 3dWarp, level 2 (MINOR), type 4 (BUG_FIX)
    Print correct obliquity transformation matrix


03 Jul 2008, RC Reynolds, plug_drawdset, level 1 (MICRO), type 4 (BUG_FIX)
    edt_dset_items.c: for .hdr, use .img brick file, storage_mode = BY_NIFTI


01 Jul 2008, RC Reynolds, to3d, level 2 (MINOR), type 4 (BUG_FIX)
    fixed crash in case of mosaic and un16, no longer having im data

    Problem found by R. McColl.


18 Jun 2008, DR Glen, align_epi_anat.py, level 2 (MINOR), type 4 (BUG_FIX)
    Bug fixes - duplicate master options, mislabeled variable


06 Jun 2008, RC Reynolds, plug_crender, level 2 (MINOR), type 4 (BUG_FIX)
    integral threshold was off by 1


29 May 2008, RC Reynolds, model_demri_3, level 1 (MICRO), type 4 (BUG_FIX)
    help update to clarify use of AFNI_MODEL_D3_R1I_DSET


22 May 2008, RC Reynolds, 3dTshift, level 2 (MINOR), type 4 (BUG_FIX)
    with -rlt, slices without any time shift must still be processed

    problem noticed by Jie Huang


21 May 2008, RW Cox, afni, level 1 (MICRO), type 4 (BUG_FIX)
    Put -1dindex into '3dclust' output from Clusterize report.  Oops.

21 May 2008, RC Reynolds, model_demri_3, level 2 (MINOR), type 4 (BUG_FIX)
    fixed incorrect scaling in Cp computation

    The error was introduced on April 8, 2008.


15 May 2008, ZS Saad, 3dfim+, level 2 (MINOR), type 4 (BUG_FIX)
    Fixed memory corruption when using more than 20 regressors


14 May 2008, RC Reynolds, model_demri_3, level 2 (MINOR), type 4 (BUG_FIX)
    fixed application of decay term

14 May 2008, DR Glen, align_epi_anat.py, level 2 (MINOR), type 4 (BUG_FIX)
    1D file names for child epi data,micro changes


08 May 2008, ZS Saad, 3dsvm, level 3 (MAJOR), type 4 (BUG_FIX)
    Fixed memory corruption caused by improper declaration of combName


07 May 2008, RC Reynolds, plug_3Ddump_V2, level 2 (MINOR), type 4 (BUG_FIX)
    allow 4D datasets to be opened (so that buckets are again usable)

    PLUGIN_dset_check() now checks NVALS instead of NUM_TIMES...


01 May 2008, RC Reynolds, model_demri_3, level 2 (MINOR), type 4 (BUG_FIX)
    treat RESID_CT as Ct(t), not C(t)


30 Apr 2008, RC Reynolds, make_stim_times.py, level 1 (MICRO), type 4 (BUG_FIX)
    replaced make_stim_files with make_stim_times.py in help


24 Apr 2008, ZS Saad, SurfSmooth, level 1 (MICRO), type 4 (BUG_FIX)
    Fixed leaky SurfSmooth.

    Leak was from one of fin_float pointers in Head07
    Also found leak in THD_extract_detrended_array and a few
    more small leaks in other SUMA function. Valgrind is good.


16 Apr 2008, DR Glen, align_epi_anat.py, level 2 (MINOR), type 4 (BUG_FIX)
    multiple changes

    Naming conventions for tlrc output, generic shell compatible,
    regridding options for epi and tlrc epi output


14 Apr 2008, DR Glen, align_epi_anat.py, level 2 (MINOR), type 4 (BUG_FIX)
    minor change

    3dAllineate options are also applied to epi to anat alignment,
    so output EPI datasets get original resolution and type.


08 Apr 2008, RW Cox, 3dTfitter, level 1 (MICRO), type 4 (BUG_FIX)
    Deal with ref (LHS) vectors that are very tiny.

    Modify thd_fitter.c so that ref vectors that are very tiny compared to
    the largest one will not be included in the regression.  Per the
    unreasonable request of Rasmus Birn.

08 Apr 2008, RW Cox, 3dAllineate, level 1 (MICRO), type 4 (BUG_FIX)
    Weighting in LPC cost function

    Realized that weighting in computing the LPC was only done on the blok
    level computation of the PC.  All the blok PC values were averaged
    together without weighting.  Modified code to use sum of weights over a
    blok as a weight for its PC.  This can be turned off by setting
    environment variable AFNI_LPC_UNWTBLOK to YES (to restore the LPC
    function to its previous behavior).

08 Apr 2008, RC Reynolds, 2dImReg, level 2 (MINOR), type 4 (BUG_FIX)
    allow zero slices, passing input as result

    Choleski factorization would fail on an empty slice.  In this case,
    return the input slices as the result (instead of crashing).
    Done with D Glen.


28 Mar 2008, DR Glen, 3dDTeig, level 2 (MINOR), type 4 (BUG_FIX)
    small negative eigenvalues are set to zero as in 3dDWItoDT

    This fix avoids FA being set to 0 for those affected voxels


26 Mar 2008, ZS Saad, 3dvolreg, level 2 (MINOR), type 4 (BUG_FIX)
    3dvolreg's -1Dmatrix_save was not always accounting for centers of rotation

    Now the output of -1Dmatrix_save always matches what one would get for
    the first sub-brick from
       cat_matvec FRED+orig::VOLREG_MATVEC_000000 -I 

26 Mar 2008, DR Glen, python, level 2 (MINOR), type 4 (BUG_FIX)
    repaired support for dry_run mode in python scripts

26 Mar 2008, DR Glen, matrix.c, level 1 (MICRO), type 4 (BUG_FIX)
    freed matrix used in matrix_sqrt functions


25 Mar 2008, ZS Saad, suma, level 2 (MINOR), type 4 (BUG_FIX)
    ROIs drawn on different surfaces now handled properly

25 Mar 2008, RW Cox, 3dDetrend, level 2 (MINOR), type 4 (BUG_FIX)
    -normalize didn't work -- what the hellll was I thinking?

    Also, added some help to explain how to use 3dDetrend on 1D files.

25 Mar 2008, RW Cox, 1dnorm, level 2 (MINOR), type 4 (BUG_FIX)
    Forgot the sqrt()!  kudos to David Perlman.


24 Mar 2008, RC Reynolds, @Align_Centers, level 2 (MINOR), type 4 (BUG_FIX)
    applied proper follower dataset orientation and floating point shifts

    The shift applied to the child datasets was based on the parent's
    orientation.  The shifts were also being truncated to integers.
    Changed with D Glen.


12 Mar 2008, ZS Saad, suma, level 1 (MICRO), type 4 (BUG_FIX)
    Changed crash in SurfaceMetrics when -spec and -i options are mixed. 


10 Mar 2008, RC Reynolds, SUMA_SurfMeasures, level 3 (MAJOR), type 4 (BUG_FIX)
    averages did not include nodes lost to -cmask

    Noticed by M Beauchamp.


07 Mar 2008, RC Reynolds, make_stim_times.py, level 2 (MINOR), type 4 (BUG_FIX)
    properly ignore empty lines, and exit on short files


05 Mar 2008, RC Reynolds, GIFTI, level 2 (MINOR), type 4 (BUG_FIX)
    fixed passing of INDEX_LIST


04 Mar 2008, RW Cox, 3dDeconvolve, level 2 (MINOR), type 4 (BUG_FIX)
    couple of small changes to help the hapless users

    * add more informative error message if allocate_memory() fails
    * force '-float' if any output prefix is NIfTI-1 format (.nii)


27 Feb 2008, ZS Saad, suma-general, level 1 (MICRO), type 4 (BUG_FIX)
    Another bout of initialization and leak fixes based on valgrind's output

    The hope is that some weird X behavior is caused by uninitialized
    variables.

27 Feb 2008, RC Reynolds, afni_proc.py, level 2 (MINOR), type 4 (BUG_FIX)
    fixed -regress_use_stim_files typo (was -regress_use_stim_times)


25 Feb 2008, RC Reynolds, plug_vol2surf, level 3 (MAJOR), type 4 (BUG_FIX)
    fixed application of cluster for sending data to suma

    Previously, clustering was only applied when the Olay and Thr sub-bricks
    were the same.


29 Jan 2008, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Fixed bug in 3dFDR -new handling of mask -- ooooppssssie.
    * And bug in FDR curves generated from signed statistics (like t).


28 Jan 2008, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Added PRINT_COMPILE_DATE macro to mrilib.h and to a lot of programs'
       -help output.
    * Fixed 'rule of 15' bug in afni_cluster.c, where the initial display only
       showed a max of 15 widget rows, even if there were more clusters, but
       the number of rows would be reset correctly on any re-clusterizing.
       Problem: didn't initialize maxclu_default at the right location.
    * Fixed problem with display of clusterized dataset which is both the
       Underlay and Overlay: the Underlay was being shown as the edited
       volume, but that looks real bad.  Solution: disable vedit feature
       in AFNI_dataset_slice() when calling from the 'get anat image' place
       in afni.c, and then re-enable it right afterwards.


16 Jan 2008, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Fix clusterize so that too small a value of vmul means just set the
       min cluster volume to 2 voxels.
    * Unfixed bug: display of clusterized func+tlrc.BRIK when func+tlrc datase
    t
       is warped from func+orig -- must fix afni_warp.c in this case.


05 Nov 2007, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Insert some sleeping to try to avoid X11 errors that have been reported
       (but not seen by me) about XGeometry.
    * If DONT_USE_XTDESTROY is #define-d, then XtDestroyWidget() is replaced b
    y
       XtUnrealizeWidget() -- this is used in the Makefile for linux_xorg7,
       where the old bug in that Xt library function seems to have risen
       from the dead.


16 Oct 2007, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Small bug in decode_linebuf() in mri_read.c -- if a non-number character
       is encountered in non-slowmo mode, infinite loop ensues and that's bad.
       Now, such an encounter leads to an immediate break out of the loop.
    * Small changes to approximate_number_string() and it's usage in 3dDeconvo
    lve.
    * Fix to afni_graph.c so that 'Boxes' plot mode turns off 'Double Plot'.


11 Sep 2007, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Modified 3dBlurToFWHM.c to remove scale factors from 'outset' if 'inset'
       had them.  Oopsie.


16 Aug 2007, RW Cox, Miscellaneous, level 1 (MICRO), type 4 (BUG_FIX)
    Older History stuff

    * Modify 3dDeconvolve.c to check -stim_times files for 0-1 inputs and for
       duplicate times.


23 Sep 2005, G Chen, 3dANOVA2 and 3dANOVA3, level 4 (SUPER), type 4 (BUG_FIX)
    Significant modifications were made to avoid statistics inflation for 
general linear contrasts when coefficients don't add up to 0. 

    See more details on: http://afni.nimh.nih.gov/sscc/gangc/ANOVA_Mod.html