path: root/noao/twodspec/apextract/Revisions

.help revisions Jun88 noao.twodspec.apextract
.nf

approfile.x
     When an aperture goes off the edge of an image there was an error
     which allowed the imio data buffer from the image to go out of bounds.
     (3/12/13, MJF)

aptrace.x
     The line1/line2 variables weren't being initialized to zero in the
     ap_ctrace() procedure.  This woule lead to old values from a previous
     run of the task being reused.  (2/6/13, MJF)

=======
V2.16.1
=======

approfile.x
     Fixed a bug in the Marsh algorithm causing segfaults on 64-bit
     platforms (Buglog 583)  (3/5/12, Valdes)

apall.par
     Changed the default for maxsep from 1000 to 100000.  This is because
     the default is when the user doesn't want to skip apertures and it is
     strange when the id jumps in the (rare) case that two apertures are
     marked with a separation of more than 1000.  (2/17/09, Valdes)

apedit.x
    The 's' key now works on the current aperture rather than the nearest.
    (10/7/08, Valdes)

apcolon.x
    The "all" mode was missing with :center.  (10/7/08, Valdes)

apall1.par
apfit1.par
apnoise1.par
apnorm1.par
apscat1.par
apparams.dat
    When the apertures parameter was added in 1996 the :apertures and
    :parameters commands were broken because of missing references in
    the associated hidden psets.  (10/7/08, Valdes)

=======
V2.14.1
=======

========
V2.12.2a
========

apextract.x
    When using APFIT to output the difference the IMIO buffer which was
    assumed to be static was invalidated because of I/O needed to create
    the difference.  A special case was added to handle this case.
    (7/7/04, Valdes)

apcveval.x	+
apedit.x
apextract.x
apfind.x
apfindnew.x
apfit.x
apgmark.x
apgscur.x
apmask.x
apnoise.x
apprint.x
approfile.x
aprecenter.x
apresize.x
apskyeval.x
apupdate.x
apvalues.x
apvariance.x
apnearest.x
apscatter.x
aptrace.x
mkpkg
    Added an interface routine to CVEVAL to avoid calling it with
    independent variables outside the range of the fit.  The fit range
    may be short because of tracing problems.  So the profile shift
    is now extended from the end points of the fitted range.
    (5/21/04, Valdes)

apupdate.x
apdefault.x
apfind.x
aptrace.x
apedit.x
apcolon.x
    Modified to check for inappropriate INDEF values in the "lower"
    and "upper" aperture settings.  (3/26/04, Valdes)

=======
V2.12.2
=======

apextract.x
    The edge weighting interpolation buffer space, interpbuf,  was
    increased by one pixel.  This makes the data buffer wider so that
    interpolation avoids using boundary extension except in cases where the
    aperture actually approaches the image edge.  Note that this change
    results in an improvement in the extracted spectra over the previous
    release where wraparound boundary extension was used.  This also means
    extractions will not be identical between the versions.
    (1/23/04, Valdes)

apextract.x
    The new routine ap_asifit was not correct.  (1/22/04, Valdes)

apextract.x
approfile.x
apvariance.x
    A problem related to the change of 10/21/03 is when the trace goes
    far outside the data buffer.  This could result in the number of points
    specified for asifit being too small for the interpolation function.
    An interface routine, ap_asifit, was added to do all the checks related
    to using asifit for evaluating the edge pixels.  (12/10/03, Valdes)

===========
V2.12.2BETA
===========

apextract.x
    The output name based on the input name for multiextension images
    produces a multiextension output with the same extension description.
    (12/3/03, Valdes)

apgetim.x
    1.	Restored ability to use image sections which was lost in the change
	on 7/13/98 for V2.11.2.
    2.	Support for multiextension data was added.  This consists of using
	a standard name based on EXTNAME and EXTVER and without the
	file type extension.  Note that this means that image names specified
	by index will be converted to extension name and extension version.
    (12/3/03, Valdes)

apextract.x
approfile.x
apvariance.x
    The call to asifit was specifying too many points to fit, a whole line
    in the data buffer, because the data vector may be offset from the first
    column of the data buffer.  This could cause a segmentation violation.
    (10/21/03, Valdes)


apwidth.cl
    Script to compute aperture widths from database.  This was written
    for a user and is saved here though it is not currently defined by
    default.  (12/2/02, Valdes)

apflat1.par
    The indirect reference needs to be spelled out without abbreviation to
    be "apflatten" instead of "apflat".  (11/18/02, Valdes)

=======
V2.12.1
=======

apextract.x
approfile.x
apvariance.x
    Modified to handle edge pixels by interpolation.  (6/19/02, Valdes)

=====
V2.12
=====

apgraph.x
apgmark.x
    When there is just one aperture the background regions are marked
    in apedit and in plotfile output.  (9/21/01, Valdes)

doc/apall.hlp
    The help page was indicating the extra information output was the
    variance rather than the sigma.  (8/19/00, Valdes)

apextract.x
    The checking for the maximum number of apertures that fit in the allocated
    memory (the "do j = i, napsex" loop) was incorrect because i was used
    instead of the loop index j.  (3/20/00, Valdes)

=========
V2.11.3p1
=========
=======
V2.11.3
=======

approfile.x
    In the previous change to the weights in the Horne algorithm
    the behavior when all data is rejected (say because the background
    is set wrong and all data is below the background) the weights
    would be set to MAX_REAL/10 which would cause CVFIT to fail.
    (3/13/00, Valdes)

apall1.par
apdebug.par
apfit1.par
apnoise1.par
apnorm1.par
apparams.par
    Reduced the polysep parameter.  (1/26/00, Valdes)

apextractbak.x	-
mkpkg
    Removed this second old copy which was accidentally introduced into
    mkpkg as well resulting in multiple copies of the same procedures
    in the library in V2.11.3beta.  (12/9/99, Valdes)

doc/apflatten.hlp
    Removed extraneous parameters not actually in task parameter set.
    (10/21/99, Valdes)

mkpkg
    Added missing dependencies.  (10/11/99, Valdes)

=======
V2.11.2
=======

apextract.x
    Added a keyword SUBAP when using echelle output with subapertures.
    (3/26/99, Valdes)

apflatten.par
apflat1.par
    Removed background subtraction as an option.  (12/11/98, Valdes)

apskyeval.x
    If the background sample region does not have explicit regions then
    the xmin/xmax region is used for the background.  (12/11/98, Valdes)

apfit.x
    Added errchk for ic_fit and ic_gfit.  (12/8/98, Valdes)

apflat1.par
    Reduced the polysep because it can go wrong.  (12/8/98, Valdes)

apextract.x
    Added check to fix cases with the lower and upper aperture limits
    are reversed.
    (7/13/98, Valdes)

apgetim.x
    Changed to call xtools routine that strips extensions.
    (7/13/98, Valdes)

approfile.x
    In the Horne algorithm the weights for rejected points were set to zero
    to eliminate them from the fit.  But if large regions are rejected
    this leaves the fit unconstrained and can lead to bad results.  A
    change was made to set the weights for the rejected points to 1/10
    of the minimum weight for the good data.
    (2/6/98, Valdes)

apextract.x
    For "echelle" format with "extras" the header was not setup properly
    resulting in WCSDIM=2 instead of 3.  (2/6/98, Valdes)

apids.x
    1.  Fix bug in IDS structure which pointed outside of allocated memory.
    2.  Variables ra/dec in ap_gids were being used both as pointers and
	double.  The ra/dec pointer usage was removed.
    3.	The realloc step at the end of ap_gids had the wrong check so it
	would never be done.
    (1/13/98, Valdes)

=======
V2.11.1
=======
=====
V2.11
=====

doc/apsum.hlp
doc/apsum.hlp
    Added missing task name in revisions section.  (4/22/97, Valdes)

apextract.x
    Removed calls to impl from inside amove in order to error check them.
    (1/24/97, Valdes)

t_apall.x
    1.  Added errchk for ap_dbwrite.
    2.  Made writing the aplast file optional.
    3.  It is a warning if the plot file can't be written.
    (1/24/97, Valdes)

apextract.x
    The step where the data is multplied by the gain was multiplying
    outside the data if dispaxis=1.  If the there are enough apertures
    and the aperture widths decrease then it is possible to get
    mulitplications of gain**naps which can cause a floating overflow.
    This was fixed.  (11/12/96, Valdes)

apertures.h
t_apall.x
aptrace.x
apresize.x
apalloc.x
apextract.x
apselect.x
aprecenter.x
apall.par
apall1.par
apfit1.par
apflat1.par
apnorm1.par
apparams.par
apnoise1.par
apdebug.par
apfit.par
apflatten.par
apmask.par
apnoise.par
apnormalize.par
apresize.par
apscatter.par
apsum.par
aptrace.par
apedit.par
apfind.par
aprecenter.par
mkpkg
doc/apextract.hlp
doc/apextras.hlp
doc/apedit.hlp
doc/apall.hlp
doc/apfind.hlp
doc/apresize.hlp
doc/apsum.hlp
doc/aptrace.hlp
doc/apfit.hlp
doc/apflatten.hlp
doc/apmask.hlp
doc/apnoise.hlp
doc/apnormalize.hlp
doc/aprecenter.hlp
doc/apscatter.hlp
    Added a new parameter "apertures" to select a subset of the apertures
    to resize, recenter, trace, extract, etc.  The parameter "apertures"
    which applied to the recentering was changed to "aprecenter".
    (9/5/96, Valdes)

apedit.key
    Alphabetized the summary command lists.  (9/3/96, Valdes)

apextract.x
    1. Onedspec output format is now allowed when nsubaps is greater than 1.
    2. Echelle format outputs all orders for a each subapertures in
       separate files.
    (4/3/96, Valdes)

apmw.x
apextract.x
    The WCS for strip extraction was wrong.  (1/31/96, Valdes)

apmw.x
    An error in computing the WCS transformation now prints a more informative
    message indicating the final extracted spectrum will be in pixel units.
    (1/4/96, Valdes)

apids.x
apedit.x
    Changed the behavior of the 'i' and 'o' keys with regard to the beam
    numbers.  These keys will now assign a beam number from the apid table
    for the selected aperture as well as all other apertures.  Previously
    the beam number was not changed which resulted in a new aperture
    number with a beam number that did not agree with the apid table.
    (10/27/95, Valdes)

doc/apextras.hlp +
apextract.men
apextract.hd
    Added a help topic on the "extras" information.  (9/5/95, Valdes)

apimmap.x
    If the image header dispersion axis is unreasonable a warning is
    printed and the "dispaxis" parameter is used instead.  (8/2/95, Valdes)

apids.x
apmw.x
doc/apall.hlp
doc/apdefault.hlp
doc/apedit.hlp
doc/apfind.hlp
    Modified to allow aperture ID table to be from an image header
    under the keywords SLFIBnnn.  The extracted image will have
    these keywords deleted.  (7/25/95, Valdes)

=======
V2.10.4
=======

apscatter.x
    When not smoothing along the dispersion there was a bug that when
    the number of points being fit across the disperision changed
    ICFIT was not reset causing an error "Range descriptor undefined".
    The routine now reset the "new" flag when the number of points
    changes.  (5/3/95, Valdes)

t_apall.x
    Using the same input and output image name in APSCATTER was still
    not right.  (2/24/95, Valdes)

apscatter.x
    Made the output image datatype be at least real.  (2/23/95, Valdes)

apextract.x
    For normalization the weights were not forced causing the gain to
    default to 0.  The change of 12/31/94 also fixed this by setting the
    gain to 1.  The file was touched but not changed.  (1/27/95, Valdes)

apextract.x
apskyeval.x
    Made the query for the readnoise only occur if needed.  Previously
    the query was made in the sky step even if the sky error estimate
    was not needed.  (12/31/94, Valdes)

apedit.x
    Needed to set clobber and review options so they are queried during
    interactive extraction.  (10/28/94, Valdes)

apimmap.x
apgetim.x
apgetdata.x
apextract.x
apmw.x
    1.	If a 3D image is given then a warning is printed and the first plane
	is used and the other planes are ignored.
    2.  Various fixes to allow image sections to be used.
    (10/12/94, Valdes)

aptrace.x
    An uninitialized memory problem was fixed.  (9/19/94, Valdes/Zarate)

apicset.x
    Fixed type mismatch in min/max function calls.  (6/13/94, Valdes/Zarate)

apextract.x
    Changed BANDID name for raw spectrum to "raw".  (5/3/94, Valdes)

apvariance.x
doc/apvariance.hlp
doc/apall.hlp
doc/apsum.hlp
    The correction to bring the weighted and unweighted total fluxes to the
    same value (called the bias factor) can produce odd values in special
    cases; such as slitlets where only part of the image contains real
    spectrum.  This could result in variance spectra with flux scaling
    errors to the extreme of a negative (inverted) spectrum.  The bias
    factor is now logged.  If the two total fluxes differ by more than a
    factor of 2 a warning (which always appears on the standard output) is
    given with the fluxes and the bias factor.  If the bias factor is
    negative a warning is given and the bias factor is ignored.
    (5/1/94, Valdes)

doc/aptrace.hlp
    Fixed typo in description of Legendre basis functions.  (4/1/94, Valdes)

apextract.x
doc/apextract.hlp
    Added output BANDID keywords to document the various output data.
    (2/4/94, Valdes)

apnoise.x	+
apnoise1.par	+
apnoise.par	+
apnoise.key	+
doc/apnoise.hlp	+
apextract.x
t_apall.x
x_apextract.x
apextract.hd
apextract.men
apextract.cl
mkpkg
    A new task for computing the noise sigma as a function of data value
    was added.  This allows checking the noise model parameter and
    can be used as a diagnostic of the profile modeling.  (8/28/93, Valdes)

apfit.x
apextract.x
    There were some additional problems with gain parameter dependencies
    in the difference, fit, and normalization output functions.
    (8/27/93, Valdes)

apgetdata.x
aptrace.x
apedit.par
apfind.par
apfit.par
apflatten.par
apmask.par
apnormalize.par
aprecenter.par
apresize.par
apscatter.par
apsum.par
aptrace.par
apall.par
apall.hlp
apedit.hlp
apfind.hlp
apflatten.hlp
apmask.hlp
apfit.hlp
apnormalize.hlp
aprecenter.hlp
apresize.hlp
apscatter.hlp
apsum.hlp
aptrace.hlp
    The nsum parameter may be negative to select a median rather than
    a sum of lines/columns.  The parameter files had to be modified to
    remove the minimum range limit and the help files modified to
    document the new option.  (8/10/93, Valdes)

===========
V2.10.3beta
===========

doc/apall.hlp
doc/apsum.hlp
    The format parameter description was added.  (6/24/93, Valdes)

apfind.x
apfindnew.x
    Removed the threshold in peak finding requiring peaks to be above zero.
    This works with the change to center1d to allow finding of apertures
    when the data is negative.  (5/5/93, Valdes)

apfit.x
    Added CCDMEAN=1. to output image in the normalization, flattening routines.
    (4/16/93, Valdes)

apextract.par
*.par
apgetim.x
    Moved the "dispaxis" parameter to a package paraemter.
    (3/8/93, Valdes)

approfile.x
    The profile was not cleared when saturated pixels are found.  This
    could cause NaNs to get into the data with the result that
    cvfit could produce garbage.
    (3/4/93, Valdes)

debug.par
apparams.par
apnorm1.par
apflat1.par
apfit1.par
apall1.par
    1.  Changed the default "fit2d" polynomial parameters to polyorder=10,
	polysep=0.95.
    2.  Changed the default "niterate" from 2 to 5.
    (3/3/93, Valdes)

approfile.x
doc/approfiles.hlp
    For the "fit1d" algorithm I doubled the order it uses to fit
    parallel to the disperison.  The order is still computed based
    on the tilt of the spectrum and the order used for the tracing
    but now that number is doubled.  (2/26/93, Valdes)

apscatter.x
    Revised the algorithm to keep the cross-dispersion fits in memory
    rather than writing them to disk as an image.  This speeds things
    up in the case of slow I/O.  (2/23/93, Valdes)

t_apall.x
apscatter.x
    1.  The temporary file name was not being passed to apscatter by
	t_apall resulting in use of the name ".imh" which is hidden.
    2.  Increased the column buffering size from 512*100 to 500000.
    (2/5/93, Valdes)

apextract.x
    imaccf was being used as a boolean when it should be an int.
    (12/13/92, Valdes)

debug.par
    A DPAR parameter file for use with debugging.  (1/12/92, Valdes)

apmw.x
apextract.x
    Rewrote this to allow extractions of an arbitrary number of apertures.
    Previously this was limited by MWCS.  The output format is now
    EQUISPEC. (1/12/92, Valdes)

aprecenter.x
    This routine was incorrectly selecting the apertures to be used.
    is_in_range (Memi[ranges], i) --> is_in_range (Memi[ranges], AP_ID(aps[i]))
    (1/8/93, Valdes and Hill)

doc/apbackground.hlp
    In responding to a concern about the 'b' key showing a fit even though
    the background type was "median" I added a paragraph explaining this.
    (1/8/93, Valdes)

t_apall.x
    The dispersion smoothing was turned off in noninteractive mode regardless
    of the task parameter.  This has been fixed.
    (12/8/92, Valdes)

t_apall.x
    Added error check for ap_plot.
    (10/14/92, Valdes)

apextract.x
    1. When the dispersion axis is 1 the data buffer may contain garbage
       because of using a malloc and because not all of this buffer is
       necessarily used.  Later the multiplication by the gain can cause
       an arithmetic exception.  The mallocs were replaced by callocs.
    2. Added errchks for the impl[123]r routines.
    (9/10/92, Valdes)

mkpkg
apextract.x
apmw.x		+
    1. Separated out the MWCS routines into another file.
    2. Added an apmw_saveim procedure to produce simple 1D format as is done
       in the ONEDSPEC package.
    (8/24/92, Valdes)

apskyeval.x
    When doing the fitted background variable roundoff among machines led
    to using different background points and, hence, gave noticibly different
    results.  The background points are now rounded to the nearest 1000th of
    a pixel which will produce the same background points on all machines.
    (8/19/92, Valdes)

apnorm1.x
    Added missing t_nlost parameter.  (8/10/92, Valdes)

aptrace.x
    There was an incorrect order in checking for failed traces which ends
    up referencing uninitialized memory.  This bug has been there for a
    long time (V2.8-V2.10) but only showed up during testing on the
    SGI port.  (7/31/92, Valdes)

The following set of changes concern the treatment of the background sample
regions and min/max fitting limits.  There was also a change in ICFIT
to check the min/max fitting limits and increase them if the sample region
is extended beyond the initial fitting limits.

--------

apdefault.x
	Now calls AP_ICSET with the image limits rather than the aperture
	limits as required by the change to that routine.  (7/30/92, Valdes)

apedit.x
    1.  The default aperture is reset after a colon command just in case
        one of the default aperture parameters has changed.  This is
        slightly inefficient but the alternative is a more complex
        AP_COLON to determine if a parameter relates to the default
        aperture.
    2.  Now calls AP_ICSET after fitting to apply the constraint that the
        fitting limits pass under the aperture.  (7/30/92, Valdes)

apicset.x
    1.  The input background limits for a new aperture (called
	by AP_DEFAULT) are now the maximum limits defined by the image size
	rather than the minimum limits defined by the aperture.
    2.  If a null default sample is given it is mapped to "*".
    3.  A sample with "*" will map to the maximum limits.
    4.  Allow the input and output pointers to be the same in order
	for the constraint that the fitting region pass under the
	aperture can be applied.  (7/30/92, Valdes)

--------

doc/apall.hlp doc/apsum.hlp doc/apbackground.hlp
    The documentation of the "median" and "minimum" options for the
    background parameter needed to be added.  (7/14/92, Valdes)

apextract.x
    The profile array needed to be corrected for the gain.  This makes a
    difference for the output formats that use the fitted profile
    (tasks APFLATTEN, APFIT: formats fit, ratio, diff, flat).
    (7/9/92, Valdes)

apfit1.par
    Was missing t_nlost.  (7/9/92, Valdes)

apparams.par
apnorm1.par
apflat1.par
apfit1.par
apall1.par
    Replace prompt pfiT with pfit as it should be.  (7/9/92, Valdes)

=======
V2.10.2
=======

apextract.x
    The WCS for the 3D images was changed to produce a WCSDIM of 3.
    (6/30/92, Valdes)

=======
V2.10.1
=======

=======
V2.10.0
=======

apextract.x
    1. The axis array which was set by a data statement was actually
       modified in the routine causing an improper WCS type to be
       set.
    2. If no coordinate label and/or units are found they are now set based
       on DC-FLAG.  Before calibrated long slit spectra ended up with the
       right wavelength coordinates but a label of Pixel and no units.
       (5/20/92, Valdes)

apall1.par
apparams.par
apfit1.par
apflat1.par
apnorm1.par
    The e_profile prompt contained a new line.  This was removed.
    (5/18/92, Valdes)

doc/apexv210.ms +
doc/revisions.v3.ms -
    Revisions summary document.  (5/11/92, Valdes)

apcolon.x
apedit.key
doc/apedit.hlp
    Added t_nlost to the list of colon commands.  (5/11/92, Valdes)

apgetim.x
    Added the qp and pl extensions to those stripped.  (5/8/92, Valdes)

apextract.x
    Added error checking such that if there is a problem with reading the
    input image WCS and warning is printed and pixel coordinates are set
    in the output image.  (5/8/92, Valdes)

=====
V2.10
=====

apextract.x
    For a single aperture using MULTISPEC WCS a dummy axis mapping was added
    to make the image appear to be the first line of a parent 2D image.
    (4/27/92, Valdes)

approfile.x
    Added a maximum order for the aphorne fitting function.  (4/24/92, Valdes)

apextract.x
    Added a error check trap to clean up and free memory in case of an
    error.  (4/24/92, Valdes)

t_apall.x
apdb.x
apedit.x
apcolon.x
apfind.x
apfindnew.x
apertures.h
    Made the number of apertures dynamic.  There is no longer a maximum
    number of apertures allowed.  (3/18/92, Valdes)

t_apall.x
    Length of format string declared as SZ_FNAME but used as SZ_LINE.
    Change declaration to SZ_LINE.  (3/12/92, MJF)

apextract.x
    Now the output extension is added only if the output name is the
    same as the input name.  Thus, if someone used <image>.ms as an
    output name they won't get <image>.ms.ms.  (2/12/92, Valdes)

apvariance.x
apskyeval.x
approfile.x
apextract.x
appars.x
    Trapped errors from getting the read noise and gain from the image header
    to produce a meaningful warning message.  (2/10/92, Valdes)

apedit.x
apfind.x
t_apall.x
apids.x
    1. Added code to ignore negative beam numbers during extraction.
    2. Negative beam numbers are only generated if an explicit assignment
       is made in the aperture id table or with 'j'; i.e. beam numbers
       generated by adding or subtracting will not have a negative beam.
    (2/10/92, Valdes)

apids.x
apedit.x
    Modified to not allow apertures numbers < 1.  (1/22/92, Valdes)

t_apall.x
x_apextract.x
    Added new entry point, apslitproc, for the slit processing tasks.
    (1/15/92, Valdes)

apfind.x
apall.par
apfind.par
    If nfind < 0 then the specified number of evenly spaced apertures are
    defined.  (1/14/92, Valdes)

apextract.x
apsum.par
apnormalize.par
apflatten.par
apfit.par
apall.par
apparams.par
apnorm1.par
apflat1.par
apfit1.par
apall1.par
doc/apsum.hlp
doc/approfiles.hlp
doc/apnormalize.hlp
doc/apflatten.hlp
doc/apfit.hlp
doc/apall.hlp
    1.  Replaced the "maxtilt" criteria for choosing the profile fitting
    algorithm with an explicit "pfit" parameter.
    2.  The parameter files were modified to remove "maxtilt", add
    "pfit", and change the default "polyorder" parameter for the
    fit2d algorithm from 4 to 6.
    3.  The "pfit" parameter is redirected from the hidden parameter files
    to the user parameter files allowing the users to select the profile
    fitting algorithm.
    (1/8/92, Valdes)

t_apall.x
apdb.x
    1.  Added special strings for the reference parameter.  If the reference
    parameter is "OLD" then only input images with existing database
    entries are processed.  If the reference is "NEW" then only input
    images without existing databse entries are processed.
    2.  Added a new procedure, ap_dbaccess, to simply check for the presence
    of a database file.  This is used for the above change.
    (1/2/92, Valdes)

aptrace.x
apall.par
aptrace.par
apparams.par
apall1.par
doc/apall.hlp
doc/aptrace.hlp
    A new parameter has been added to set the number of steps which may
    be lost during tracing.  (9/5/91, Valdes)

apextract.x
    Changed ap_setdisp to ap_wcs.  This routine uses MWCS and sets the
    WCS to multispec.  (8/29/91, Valdes)

apextract.x
    Modified so that profile image is used in place of input image for
    determining the profile and eliminated the use of a disk profile file.
    This is inefficient if the same profile image is used for many input
    images but it is much easier for the user to understand.
    (8/27/91, Valdes)

apvariance.x
    The subaperture extraction did not work because of a typo.
    (8/27/91, Valdes)

apextract.x
    The profile image capability had several bugs which were fixed.
    (8/27/91, Valdes)

approfile.x
    Fixed another division by zero problem in ap_horne.  (8/27/91, Valdes)

approfile.x
    Failed to clear profile in the case the spectrum was negative.
    (5/30/91, Valdes)

apertures.h
    Increased the maximum number of aperture from 100 to 1000.
    (4/29/91, Valdes)

apdefault.x
apicset.x
    The default aperture was setting the background fitting range to
    the full image range rather than range covered by the sample region.
    This could cause singular solution errors in some cases.
    (3/27/91, Valdes)

apfind.x
    Allowed still finding the specified number if some apertures (up to
    2) fail for some reason such as too near the edge.  This is done by
    setting the number of candidates to nfind+2. (3/26/91, Valdes)

aprecenter.x
doc/aprecenter.hlp
    When using the shift option the shift is now the median (including
    averaging of central 2 shifts for even number of peaks) instead of
    the average.  (3/26/91, Valdes)

appars.x
apall1.par
apfit1.par
apflat1.par
apnorm1.par
apparams.par
    In order to allow writing to redirected parameters rather than overwrite
    the redirection string a kluge was added using the prompt string.
    The apput procedures check the prompt string for the first character
    ">" and if present write to the parameter given in the rest of the
    string.  All the hidden parameter files with redirected parameters
    had to be changes.  (3/26/91, Valdes)

apextract.par
apall.par apall1.par
apsum.par apdefault.par apparams.par
apfit.par apfit1.par
apflatten.par apflatten1.par
apnormalize.par apnorm1.par
    1.  Moved format parameter from package parameters to
	APALL and APSUM.
    2.  Moved dispaxis parameter from package parameters to APDEFAULT.
    3.  Added new background types.
    (3/21/91, Valdes)

t_apall.x
    Allow scattered light correction to be run by APSCRIPT.
    (3/21/91, Valdes)

apscatter.x
    Moved CLIO call for anssmooth out of loop to a single call using a
    procedure variable.
    (3/21/91, Valdes)

apextract.x
apskyeval.x
    Aded new background functions median and minimum.
    (3/21/91, Valdes)

=================================
V3 of APEXTRACT Installed 8/23/90
=================================

apextract$exsum.x
    Added a test for the existence of output images before extractions and
    a query to select whether to clobber spectra or not.  (8/9/89, Valdes)

apextract$exmvsum.x
apextract$exfit.x
    1.  The moving average for profile images did not work correctly.  It
	subtracted the line moving out of the moving average but failed to
	add in the line moving into the average.  This caused the profile
	to depart more and more from the data as the extraction procedes
	through the image.
    2.  The moving average for profile images actually used naverage + 1
	instead of naverage becuase in this case the line being extract is
	also included in the average.
    3.  The fitting of the model to the data had a funny behavior when the
	model had negative values and variance weight was used.  The
	negative values are now excluded from the fitting calculation in
	this case.  Note that the model is supposed to not contain negative
	values.  Also note that if variance weighting is not used then the
	negative values are used in straight least-squares fitting.
	(8/8/89, Valdes)

apextract$apicset.x
apextract$apedit.x
apextract$apupdate.x
    The region over which the background fitting is defined has been extended
    to require overlapping the aperture. (7/31/89, Valdes)

apextract$apcolon.x
    Added gdeactivate/greactivate calls when using EPARAM.  (6/14/89, Valdes)

apextract$apextract.x
    If reference apertures are used without change (no recenter, edit, or
    retrace) then the apertures were not written to the database.  This has
    been changed to write (or query if interactive) the apertures if
    dbwrite=yes.  (5/19/89, Valdes)

apextract$exio.h
apextract$exio.x
apextract$exsum.x
    1.  The maximum number of simultaneous extractions was increased from
	20 to 50.
    2.  The amount of buffering used for column access (dispaxis=1) was
	increased from 100K pixels to 1M chars.  (5/12/89, Valdes)

apextract$doc/apsum.hlp
    Added a sentence about the proper alignment of echelle spectra.
    (5/8/89, Valdes)

apextract$t_apscatter.x +
apextract$apscatter.par +
apextract$apscat1.par +
apextract$apscat2.par +
apextract$doc/apscatter.hlp +
apextract$mkpkg
apextract$x_apextract.x
apextract$apextract.cl
apextract$apextract.men
apextract$apextract.hd
imred$echelle/apscatter.par +
imred$echelle/apscat1.par +
imred$echelle/apscat2.par +
imred$echelle/echelle.cl
imred$echelle/echelle.men
    Added a new task, APSCATTER, to fit and subtract scattered light.
    It includes two hidden psets, APSCAT1 and APSCAT2.  (3/3/89, Valdes)

apextract$apnormalize.par
imred$echelle/apnormalize.par
    Input image parameter name needed to be changed from "images" to "input".
    (2/28/89, Valdes)

apextract$exio.x
    Added errchk for imgeti related to getting DISPAXIS keyword which
    might be missing.  (2/28/89, Valdes)

apextract$exsum.x
    Changed usage of CRPIX keyword to real from integer.  (1/25/89, Valdes)

apextract$apgmark.x
    The size of the aperture labels now decreases with increasing number
    of apertures.  (1/24/89, Valdes)

apextract$t_apnorm.x
apextract$apnormalize.par
imred$echelle/apnormalize.par
apextract$doc/apnormalize.hlp
    Changed parameter names "lowreject" to "low_reject" and "highreject"
    to "high_reject" to be consistent with other ICFIT tasks.
    (1/24/89, Valdes)

apextract$aprecenter.x
    The progress information was not using the verbose parameter.
    (1/23/89, Valdes)

apextract$apedit.x
    Fixed bug causing 'i' info to only be visibly momentarily.
    (12/16/88 Valdes)

apextract$t_apnorm.x
    When fitting the normalization spectrum interactively any deleted
    points would be remembered in the following apertures.  Deleted points
    are now cleared after interactive fitting.  (12/15/88 Valdes)

apextract$apedit.x
    Fixed a minor bug in the 's' option (center + wx --> wx) (12/15/88 Valdes)

apextract$*.x
apextract$apio.par
apextract$doc/apio.hlp
    Removed the beep option. (12/8/88 Valdes)

apextract$exgmodaps.x
    The temporary apertures array and number of apertures were not being
    initialized correctly causing a segmentation violation every time
    the program was run. Added an n = 0 statement and initialized all the
    pointers to NULL. (10/13/88 Davis)

apextract$exmvsum.x
    When cleaning and using a moving average the replacement of the bad pixel
    in a data profile back in the moving average was in error leading to 
    bad results and possible memory corruption.  (10/9/88 Valdes)

apextract$exfit.x
apextract$exmvsum.x
    1.  The sigma clipping test now uses the variance relation for testing the
	residuals.
    2.  A bug was causing the length of the profile image to be incorrectly
	referenced leading to an out of bounds error.

apextract$apedit.x
apextract$exapsum.x
    1.  After doing a ":line" new apertures were defined with the old line
	as the aperture center.  This made tracing fail.
	Now after changing the line the default aperture is updated and
	the apeture center is explicitly set every time.
    2.  In a rare circumstance a divide by zero error occured for the fitted
	background.  This is now checked.  (7/19/88 Valdes)

noao$lib/scr/apedit.key
apextract$apedit.x
    1.  Deleted a reference to :nfind in the cursor help.
    2.  After deleting a background defintion it was no longer possible
	to define a new one.  Now an attempt to define backgrounds
        after they have been deleted is initialized to the default again.
        (7/1/88 Valdes)

apextract$exsum.x
    Fixed a bug in blocking overwriting of existing echelle format spectra.
    Fixed a minor bug in writing the aperture info to the header of a 1D
    format sky spectrum.  (6/16/88 Valdes)

apextract$apgmark.x
    Labels are not written outside of graph range.  (5/20/88 Valdes)

apextract$apedit.x
apextract$exgraph.x
noao$lib/scr/apedit.key
    Added 'I' interrupt.  (4/20/88 Valdes)

apextract$exio.x
    EX_UNMAP when using column access was free a buffer supplied by
    IMIO causing a memory corruption error on VMS.  (3/30/88)

apextract$exgmodaps.x
    Model apertures where not being match up correctly resulting in a
    warning message.  (3/22/88 Valdes)

apextract$exsum.x
apextract$exstrip.x
apextract$exrecen.x
    Added iferr statements for asifit.  Without this the task crashed
    when an aperture went off the edge.  (3/10/88 Valdes)

apextract$* Major changes:
	o Use profile template image
	o Don't interpolate data profile before cleaning
	o New background option
	o Integrate apertures
	o Restructure code
	o And much more
    (3/1/88 Valdes)

==============
APEXTRACT V2.0
==============

apextract$excextract.x -
apextract$exlextract.x -
    Removed these unused procedures. (1/5/88 Valdes)

apextract$excstrip.x
apextract$exlstrip.x
    Free curfit pointer if ic_fit returns an error. (1/5/88 Valdes)

apextract$excsum.x
apextract$exlsum.x
    Set background to zero if ic_fit returns an error. (1/5/88 Valdes)

    Original:
	iferr (call ic_fit (AP_IC(aps[i]), cv, Memr[x],
	    Memr[bufin], Memr[w], ncols, YES, YES, YES, YES))
	    ;

    New:
	iferr {
	    call ic_fit (AP_IC(aps[i]), cv, Memr[x],
	        Memr[bufin], Memr[w], ncols, YES, YES, YES, YES)
	    call ex_apbkg (cv, ncols, center, low, high,
	        skyout[line,i])
	} then
	    skyout[line,i] = 0.

apextract$exapsum.x
    Added check for no data in sum. (1/5/88 Valdes)

    In procedure ex_apsum:
	a = max (0.5, center + lower)
	b = min (npts + 0.5, center + upper)

	if (a >= b) {			<-- ADD
	    sum = 0.			<-- ADD
	    return			<-- ADD
	}

    In procedure ex_apbkg:
	a = max (0.5, center + lower)
	b = min (npts + 0.5, center + upper)

	if (a >= b) {			<-- ADD
	    bkg = 0.			<-- ADD
	    return			<-- ADD
	}

apextract$apnormalize.x
    There was no error checking on ap_dbread which caused misleading errors
    (apio package not found).  This was fixed by updating the logic to be the
    same as in apextract.x.  A quick fix for outside sites is to add an
    errchk for ap_dbread.  (12/18/87 Valdes)

apextract$apedit.x
apextract$apgscur.x
    When first entering APEDIT with apertures defined the first attempt to
    point the cursur at the first aperture uses an indefinite y value since
    no cursor read has yet taken place.  The choices are to set it to some
    arbitrary value, some percentage level on screen, or do nothing.  I
    chose to do nothing.  Thus, the cursor will not point at the current
    aperture in this case but it will leave the cursor position unchanged.
    (12/17/87 Valdes)

apextract$apio.par
apextract$exsum.x
apextract$exoutsum.x
apextract$apedit.x
    Added new parameter "format" to APIO pset.  Added new output formats for
    echelle and multispectra data consisting of a single 2D image.  Added new
    information in header.  Old format is called "onedspec".

    Removed debugging print statement.  When it was put in I don't know.
    (12/17/87 Valdes)

apextract$apextract.x
    An error was introduced with the change of 11/9/87 such that the error
    when a database file does not exists was no longer being trapped.  This
    was intended for reference images but was not intended for creating new
    database files.  An iferr was added.

apextract$t_apnormalize.x
apextract$trtrace.x
apextract$exstrip.x
apextract$exsum.x
    ERRCHK declarations for IMGETI were added to detect a missing DISPAXIS.
    Failing to catch this error caused misleading error messages and
    other fatal errors. (11/9/87 Valdes)

apextract$apextract.x
apextract$apdb.x
    An error reading apertures for a specified reference image is now
    printed instead of assuming there are no reference apertures.
    (11/9/87 Valdes)

apextract$exfit.x
    Changed the variance formula from
	V = v0 + v1 * abs (S + B)
    to
	V = v0 + v1 * max (0, S + B)		if v0 > 0
	V = v1 * max (1, S + B)			if v0 = 0
    (11/9/87 Valdes; see also 8/6/87)

apextract$t_apnormalize.x
    When normalizing more than 20 apertures some of the apertures
    were being lost in the output image.  This was fixed to only
    fill between the apertures on the first set of apertures.

    Normalizing when DISPAXIS=1 did not work.  There were a number of
    errors.  This task was never tested with data in this orientation.
    (9/22/87 Valdes)

apextract$exfit.x
apextract$apsum.par
apextract$apstrip.par
apextract$doc/apsum.hlp
apextract$doc/apstrip.hlp
    When computing the variance for doing variance weighting the
    intensity could be negative.  An absolute value was added to
    correct this.

    Make defaults for naverage=100 and nclean=2 because people have
    been using inadequate number of lines for low signal-to-noise
    data and because of interpolation even an 1 pixel cosmic ray
    expands to two pixels.  (8/6/87 Valdes)

====
V2.5
====

apextract$apio.par
apextract$apio.x
apextract$t_apnormalize.x
apextract$exsum.x
apextract$exstrip.x
apextract$apfindnew.x
apextract$apfind.x
apextract$apextract.x
apextract$apedit.x
apextract$trtrace.x
apextract$trltrace.x
apextract$trctrace.x
apextract$exgraph.x
apextract$doc/apio.hlp
    Valdes, May 19, 1987
    1.  Added the parameter "verbose" to the APIO pset to allow turning off
	log information to the terminal.
    2.  Made changes to minimize switching between text and graphics mode.
	With verbose=no there should be essentially no mode switching.  This
	was done for terminals in which such switches is either annoying
	or slow.

apextract$mkpkg
apextract$apcvset.x
apextract$t_apnormalize.x
apextract$trtrace.x
apextract$exsum.x
apextract$exstrip.x
apextract$apgetdata.x
apextract$apimmap.x +
    Valdes, April 24, 1987
    1.  Protection against using an image which is not 2 dimensional was added.

apextract$apedit.x
apextract$apextract.x
apextract$apsum.par
apextract$exapsum.x
apextract$excsum.x
apextract$exgprofs.x
apextract$exlsum.x
apextract$exoutsum.x
apextract$exsum.x
apextract$doc/apsum.hlp
    Valdes, April 11, 1987
    1.  Added additional option to APSUM to output the subtracted sky
	background spectra when doing background subtraction.

apextract$apextract.hd
apextract$apextract.men
apextract$apextract.par
apextract$apgetim.x
apextract$apnormalize.par
apextract$mkpkg
apextract$t_apnormalize.x +
apextract$x_apextract.x
apextract$doc/apnormalize.hlp +
    Valdes, April 3, 1987
    1.  New task APNORMALIZE installed.

apextract$*x
    Valdes, February 17, 1987
    1.  Required GIO changes.

apextract$apio.h
apextract$exgraph.x
apextract$apio.x
apextract$excextract.x
apextract$apextract.tar
apextract$trltrace.x
apextract$trctrace.x
apextract$exlextract.x
apextract$applot.x
apextract$apedit.x
    Valdes, February 13, 1987
    1.	I've made a number of modifications to the APEXTRACT package to
	correct problems in the way graphics and text are mixed.  The
	main change was to make the graphics open very local to the
	procedure doing the graphics.  Previously the graphics device
	was opened at the beginning of the logical task and remained
	open until the end.  This was done since the integrated nature
	of the package has many procedures which may do graphics.  This
	has the bad side effects that for the first graphics open of
	the process the terminal enters graphics mode (graphics screen
	on SUN and clear screen on VT640) well before any graphics is
	performed and text I/O is a problem.  Putting GOPEN and GCLOSE
	immediately around the code that does graphics fixed almost all
	the problems.  The only time that text output is now done when
	the graphics terminal is open is for '?' and ':show' type of
	commands.  The only remaining problem is the page wait problem
	associated with '?' occuring before a cursor read rather than
	immediately after the text output causing the last written
	status line to become confused with the page wait prompt.

apextract$apfind.par
    Valdes, February 12, 1987
    1.  The parameter apfind.nfind was changed back to hidden in order
	for apsum to work in the background.  Note the PSET mechanism
	would fix this by allowing nfind to be specified on the command
	line.

apextract$apio.par
    Valdes, February 9, 1987
    1.  New users rarely look at or know about the log files produced by
	the package tasks.  These files (particularly graphics) take up
	a lot of space.  Therefore the default is now not to log
	text or graphics output.

apextract$apedit.x
apextract$apsort.x
apextract$exapsum.x
apextract$trltrace.x
apextract$trctrace.x
apextract$doc/apedit.hlp
noao$lib/scr/apedit.key
    Valdes, February 2, 1987
    1.  Added a new edit option, 'o', to reorder the aperture id and beam
	numbers sequentially.  This is useful after apertures have been
	deleted and added interactively.
    2.  If the aperture went completely off the image (ie there is no
	overlap of even part of the aperture with the image) a random value
	was given the aperture sum because the value was not intialize to
	zero.  It is now initialize to zero.
    3.  There is now a requirement that more than three points be traced
	before a trace will be fit.

apextract$trltrace.x
apextract$trctrace.x
apextract$apgetdata.x
    Valdes, January 30, 1987
    1.  The middle line (default) was not calculated correctly.  
    2.  The tracing would sometimes run beyond the end of the image.  The
	value of this traced point is suspect.  It was found because there
	would be a point off the end of the ICFIT graph which was set to
	be the size of the image.
    3.  Just in case a traced point is right at the edge I made the default
	window for fitting the trace extend half a step beyond the edges
	of the image.

apextract$exsum.x
apextract$exstrip.x
    Valdes, January 30, 1987
    1.  Skip Schaller (Steward) reported a problem with running out of memory
	with an input list of 20 images.  Also when the user deleted input
	images after they were processed but before the list was finished
	(to reduce memory) the memory was not actually freed until the process
	actually finished.  I found that the input images
	where not being closed! (Sorry about this really stupid error.)
	This may not be all the problem but it is probable since IMIO
	maintains large buffers.

apextract$apcolon.x
    Valdes, January 15, 1987
    1.  Replaced dictionary string and numeric case by macros for readability.

apextract$apfind.par
    Valdes, December 19, 1986
    1.  Change the default NFIND parameter in APFIND to auto mode.

apextract$apnearest.x
    Valdes, December 12, 1986
    1.  It is possible for more than one aperture to be equidistant from
	the cursor, particularly when more than one aperture is defined for
	the same feature.  This is ambiguous for those commands which operate on
	the nearest aperture.  The modification will query the user if it
	is found that there is more than one aperture at the same distance
	from the cursor.

apextract$peaks.x
    Valdes, October 10, 1986
    1.  VMS adjustable array dimension error found.  Replaced declaration
	dimension by ARB since the passed dimension value may be zero.

apextract$*
    Valdes, September 16, 1986
    1.  A new version of the package has been installed.  It is very
	different from the old version.  The user parameter files must
	be unlearned.

====================
New Package Released
====================

apextract$: Valdes, July 19, 1986
    1.  APEDIT and TRACE modified to include a detection threshold parameter
	for profile centering.
    2.  The help pages were updated.

apextract$apedit.x, apvalue.x: Valdes, July 17, 1986
    1.  A bug was found that appears if you edit the apertures of a
	previously traced image.  The center moves with use of 'l' or
	'u'.  This is due to an inconsistency between whether the aperture
	center is before or after the traced shift is added.  Changes
	to APVALUE.X and the calls to it in APEDIT.X fix this.

apextract$: Valdes, July 15, 1986
    1.  TRACE had a bug when trying to specify an explicit starting
	line to edit and trace from.  This was fixed.
    2.  EXCEXTRACT.X, EXLEXTRACT.X, and EXGPROFS.X were modified to
	check for errors from background fitting.  This arose when
	trying to get the background for a spectrum which has gone
	off the edge of the image.

apextract$apedit.x, apsort.x : Valdes, July 9, 1986
    1.  When changing the aperture number the apertures are sorted and
	then the wrong aperture could become the current aperture if
	another aperture exists with the same aperture number.  This was
	fixed so that the sorting returns the correct current aperture
	after sorting.  Also apindex.x is no longer needed.

apextract$apedit.x: Valdes, July 7, 1986
    1.  The 'd' delete key now does nothing if no apertures are defined.
	Previously it would cause a failure of the task.

apextract$apedit.x:  Valdes, July 7, 1986
    1.  Added redraw and window commands.
    2.  Help page and '?' menu updated.

apextract$apedit.x:  Valdes, July 3, 1986
    1.  APEDIT modified to use new ICFIT package.

apextract$apgetim.x:  Valdes, July 1, 1986
    1.  New procedure to strip the image extension.  This is necessary
	to create proper database files and to avoid having two legal
	names for images in the database.

apextract$exapstrip.x:  Valdes, July 1, 1986
    1.  Simple strip extraction without profile modeling interpolated
	the data so that the lower edge of the aperture was centered
	on the first pixel.  This is inconsistent with the other extractions
	which put the center of the aperture at the center of a pixel.
	This has been fixed.

apextract:  Valdes, June 30, 1986:
    1.  TRACE was not correctly initializing the new ICFIT package.
	In particular the task parameters "function" and "order" had no
	effect and when multiple files were used the last set parameters
	were not retained.  Changes to t_trace.x, trctrace.x, trltrace.x.

=====================================
STScI Pre-release and SUN 2.3 Release
=====================================

apextract:  Valdes, June 20, 1986:
    1.  New APEXTRACT installed.  This version includes background
	subtraction and new ICFIT.

apextract:  Valdes, June 2, 1986
    1.  Another round of name changes.  EDITAPS -> APEDIT,
	EXTRACT1 -> SUMEXTRACT, EXTRACT2 -> STRIPEXTRACT.

apextract:  Valdes, May 16, 1986
    1.  Renamed APDEFINE to EDITAPS.
    2.  Moved parameters used in editing the apertures from EXTRACT1 and
	EXTRACT2.  These are now obtained from EDITAPS regardless of which
	task calls apedit.
    3.  Added parameters "cradius", "cwidth", "ctype", "lower", and "upper"
	to EDITAPS.  The first parameters control profile centering and the
	last parameters set the default aperture limits.
    1.  Added new keys.  'c' center current aperture of profile near the cursor.
	'm' mark and center aperture on profile near the cursor.  's' shift
	the center of the current aperture to the cursor position.  The change
	allows centering of profiles using CENTER1D.

apextract$extract.x:  Valdes, May 13, 1986
    1.  EXTRACT has been broken up into two tasks; EXTRACT1 and EXTRACT2.
	EXTRACT1 extracts weighted summed 1D spectra.  EXTRACT2 extracts
	2D apertures corrected for shifts across the dispersion.

apextract:  Valdes, April 23, 1986
    1.  Modified EXTRACT to only warn about an existing output image.
	This allows adding a new aperture without needing to delete
	old apertures or reextract previous extractions.

apextract:  Valdes, April 21, 1986
    1.  All procedures which pass the number of current apertures as an
	argument and then dimension the array with this number
	were modified by dimensioning the array as dimension
	APS_MAXAPS or ARB.  This was done because the number of apertures
	might be zero.  This is a fatal error on VMS/VAX though not on UNIX/VAX.

apextract:  Valdes, March 27, 1986
    1.  Modified EXTRACT (exsum.x and exwt.x) to update the dispersion image
	header parameters.  In particular if the input dispersion axis is 2
	then the header parameters must be reset to dispersion axis 1.

===========
Release 2.2
===========
.endhelp