version 1.23.x

  o Populate new eSet slot 'protocolData' with information from cel files.


version 1.17.x

  o Replaced usage of defunct Biobase classes exprSet and phenoData with
    ExpressionSet and AnnotatedDataFrame.


version 1.4.x [under development]

  o xy2indices/indices2xy  have now a (dangerous) parameter xy.offset.
    See below for more details.

  o a new package-wide parameter called 'xy.offset' was added. This is
    the first towards getting rid of the 'indexing-starting-at-one-and
    -not-at-zero' difficulty. This package-wide parameter is the only
    thing one should play play with to change the offset of x/y
    coordinates for features on a chip.

  o computeExprSet (method of AffyBatch) has the following (visible)
    improvements:
      - better reporting of errors
      - better handlings ids as parameters (does not crash any longer when
        unknown ids are given)

  o indexProbes (method of AffyBatch) sees the deprecated flag 'xy' removed
    for good.

  o mmindex (see indexProbes above)

  o pmindex (see indexProbes above)
 
  o ppsetApply: a function to apply a function over the probe sets that can be
    built from an instance of AffyBatch. This is done in the spirit of
    esetApply in Biobase (the covariate information in phenoData are
    directly accessible from the function (see example in the man page). 

  o rma - the subset parameter now works properly.

  o mva.pairs - now uses a subset to fit the loess curve (much more efficient).
    also the summary statistics displayed are different. In particular the 
    median and IQR of the M's are shown in the lower triangle. Previously,
    an IQR of loess curve values was shown.
	 

version 1.3.x 

Things done:

  0 (wh:) moved everything related to Cdf-class from 'affy' to 'makecdfenv':
    R/Cdf.R, R/getLocationsData.Cdf.R, locate.name.R, pmormm.R
    R/read.cdffile.R, data/cdf.example.rda, man/Cdf-class.Rd
    man/getLocationsData.Cdf.Rd, man/locate.name.Rd, man/pmormm.Rd
    man/read.cdffile.Rd, src/read_cdffile.c

  o read.affybatch2 reads all cel files together in one big hit
    into an affybatch

  o the baseline in normalize.AffyBatch,invariant can be selected
    using a baseline.type parameter. Also changed "progress" to the 
    more standard "verbose". Seemed to be an error since both PM and
    MM probes were normalized by only PM's where used to establish the 
    normalization, now follows same standard as been introduced for
    normalize.AffyBatch.quantiles.

  o Most normalization routines can now be called with the parameter 
    type which specifies whether to be pmonly, mmonly, both together 
    or both separately. This introduces more consistency to the how
    the normalization routines are actually applied to affybatchs.
 
  o express() has been completely removed.


Things that would be good to see before next release:
  o More consistent usage/application of MIAME
	

version 1.2.x:

  o Autoload of cdfenvs on demand (uses reposTools). Can be
  configured through the options.

  o slot 'preprocessing' of the MIAME attribute used to store
  normalization step information [will be polished, list returned
  for the moment. Need for a class and check compliance with MIAME
  standards]
  
  o default methods for normalization, bg correction, pm correction
  and summary now in the package options [options exist for all, but
  only used by normalize for the moment].

  o tuning of the MAS5.0 methods implemented (bgcorrect.mas, ...)
  [Ben for details. add URL for the comparison he made]

  o method plot.ProbeSet, an alternative to barplot, to plot probe
  level information.

  o parameter 'scale' in the method barplot for ProbeSet. All the
  barplots are scaled to eachothers.This is the default.

  o bug in the parser fixed (infinite loop reported with apparently
  non-standard CEL files.).

  o bug in the parser fixed (the 'sd' data returned were not correct).

  o missing slot in the dataset SpikeIn fixed.

  o The function express() is deprecated. It still functions normally 
  but gives warning mesage. It will be removed in a future release.
  The function expresso() should be used as a replacement.

  o bug in normalize.AffyBatch.qspline fixed (thanks to people at
  Insightful). The expression data matrix sent to normalize.qspline
  was mistakingly transposed.

  o The default background on the rma() function has been changed. Now 
  the results from rma() and expresso() should agree completely.

  o New functions 'xy2indices' and 'indices2xy' to shuttle from
  x/y pos to indices (like the ones in cdfenvs) (and reverse).

  o Reformating of the documentation.
 
version 1.1.x:

  o 'image(cel)' scales to the size (# rows and # cols) of a chip.
  locations can be plotted over directly ('plotLocations' was fixed
  accordiginly).

  o 'write.celfile(cel)' to write Cel objects into .CEL files

  o 'getInfoInAffyFile' to snoop in CEL and CDF files

  o 'whatcdf' to get the name of the CDF from the CEL

  o one more slot in class 'Cdf': cdfName (will match with what is 
  returned by 'whatcdf')

  o one more slot in class 'Cel': cdfName (returned by whatcdf)

  o new class 'AffyBatch': cdfName (returned by whatcdf)

  o extensive set of tests in the directory 'tests/'