2017-07-29:
  - Fix refactoring: `c.memisc_mtable` and `[.memisc_mtable` now return objects
    with the correct class attribute.
  - Make package depend on versio 3.4

2017-07-13:
  - Fix handling of matrices contained in data frames by `toLatex()`
  - `useDcolumn` and `useBooktabs` arguments of `toLatex()` methods now have
    global options as defaults

2017-07-11:
  - Restrict visibility of object code symbols to package initialisation routine

2017-07-10:
  - Deal with PROJECT issues reported by Tomas Kalibera: Add PROTECT to objects
    for which attributes are set
  - Provide Call method registration for functions called from C-code

2017-07-09:
  - Bugfixes for `repr_html()` methods
  - Eliminate dependence on IRdisplay package

2017-07-03:
  - Change class of objects created by `mtable()` from "mtable" to "memisc_mtable"
    to avoid name clash with objects created by `model.table()` from package
    "stats".

2017-07-02:
  - Support for output in Jupyter notebooks.

2017-06-02:
  - Completed support for output as print, LaTeX, and HTML for new mtable
    architecture.

2017-05-30:
  - Introduced support for multi-line headers.

2017-05-26:
  - Architectural change for objects produced by `mtable()`. This allows
    several types of parameters, such as coefficients and variances.

2017-05-26:
  - Adapted `show_html()` to use within Jupyter Notebooks

2017-05-19:
  - Created methods for `repr_html()` to support HTML output in Jupyter
    notebooks

2017-03-05:
  - Fixed handling of empty summaries in `mtable()`
  - Better "splash screen" section of LaTeX formatted mtable objects.

2017-02-03:
  - Fixed a bug in `relabel` method for `item'
    objects. `relabel(x,...,gsub=TRUE)` should now work as expected.
  - Fixed a bug in `readLabPorStream`, coming from wrong assumptions about the
    encoding of value labels of character string variables in "portable"
    files. Some more SPSS "portable" files should now read into R.

2016-12-24:
  - New `withVCov` and `withSE` generic functions to more
    conveniently provide sandwich estimates of variances and standard
    errors

2016-09-28:
  - More efficient 'subset' method for large external data sets

2016-07-09:
  - No 'useRs' in DESCRIPTION and software names in single quotes

2016-07-08:
  - Suggest 'rmarkdown' package

2016-07-02:
  - Fix type that crept in while fixing bug in parseSysHeader

2016-06-05:
  - Fix 'felicity' issue pointed out by Martin Maechler
  - Make formatting of codebooks a bit more robust

2016-04-30:
  - Fix issue #8 - `toLatex.ftable()` does no longer attempt to show
    titles if they are not available.
  - Fix issue #1 (error in 'parseSysHeader')

2016-04-29:
  - Make `cases()` robust with respect to NA conditions.

2016-04-17:
  - Fix extracolsep in `toLatex.ftable()` as suggested by Kirill Müller

2016-02-21:
  - Make the use of ampersand-escapes in HTML output optional and use unicode characters
    by default. (So that pandoc will not escape them.)
  - Minor improvenent of `c.mtable()`.

2016-02-07:
  - Unified headlines in vignettes.

2016-02-02:
  - Implemented "Date.item" class, which allows to use `data.set` or `as.data.set`,
    etc. with "Date" objects.

2016-01-29:
  - Use `any(is.na(...))` instead of `anyNA(...)` in `[.mtable`.

2016-01-27:
  - Fixed `show` and `codebookEntry` methods for "datetime.item" class.

2016-01-24:
  - Do not wrap HTML tables and list in HTML paragraphs in `format_html.CodebookEntry`.

2016-01-21:
  - Correct spelling and case of vignette titles.

2016-01-16:
  - Use "html_viewer" option instead of "show_html.output".

2016-01-04:
  - Deleted last sentence in package description.
  - Put back vignette for `ftable-matrix`.
  - `show_html` can be used with arbitrary output functions, e.g. `rstudio::viewer`.

2016-01-03:
  - Removed `UnZip` from package, which has been redundant for some time.
  - Use `knitr::rmarkdown` instead of `knitr::knit` in vignettes.
  - Added vignette for `ftable_matrix`.
  - `show_html` can be used with arbitrary output functions, e.g. `rstudio::viewer`.

2016-01-02:
  - Small vignette update.
  - New maintainer address

2015-12-31:
  - R-dependency should be >= 3.0, because this is needed for knitr vignettes.
  - Some improvements of ANES vignette.
  - Improvements for variable descriptions in data sets and importers, 
    including HTML formatting.
  - Added explanation of `tolower` to ANES Vignette.
  - New argument `margin` of `format_html` method functions, 
    which determines the positioning of the HTML tables. By default it centers the table horizontally.
  - Export `format_html` cell standard styles.
  - Renamed `html_style` into `css`, etc.
  - Automatic HTML output in ANES vignette.
  - Remove LaTeX/PDF Vignette
  - Added documentation for `html()` and friends. 
    Implemented `show_html()` method function for HTML code.

2015-12-30:
  - Make concatenation of html-elements safer

2015-12-29:
  - Added documentation for `toLatex` method for "ftable_matrix" objects
  - No longer make a fuss about duplicate model names in `combine_mtables`

2015-12-28:
  - Added example for named model groups in `mtables`.
  - Added documentation of `format_html.ftable_matrix`.
  - Updated documentation of `mtable_format_latex`
  - Create and use `Write` method function for `ftable_matrix` objects.
  - Added documentation of `ftable_matrix` objects

2015-12-18:
  - Added support for with model groups in `mtable`. `c.mtable` now creates 
    groups of models, if arguments are tagged.

2015-12-16:
  - Fixed issues shown by R CMD check
  - Adapt `format_html.ftable_matrix` to html-objects infrastructure.

2015-12-13:
  - Enable automatic insertion of linebreaks into HTML code.
  - Export methods for html objects etc.

2015-12-12:
  - Adapted HTML formatting of 'codebook's to new HTML infrastructure.
  - Added syntactic sugar to 'html()'. Corrected formatting of fixed-point numbers.
  - Made formatting of value filters safer.

2015-12-07:
  - Encapsulate fixed-point numbers into "html_group" objects
  - Update version number

2015-12-06:
  - Start object-oriented infrastructure for the creation of HTML code.

2015-11-13:
  - Import 'str' and 'ls.str' from 'utils' package to prevent a NOTE in R CMD check

2015-11-09:
  - Make `percentages()` a little more robust against empty cells in marginal tables.

2015-09-24:
  - Fixed Latex output of data frames.

2015-09-13:
  - Fix regressions caused by `rbind()` and `cbind()` methods for "ftable" objects.

2015-09-12:
  - Added some new `relabel()` methods.

2015-09-06:
  - Fixed treatment of value labels in HTML formatting of codebooks.

2015-09-03:
  - Added basic HTML formatting of `ftable_matrix` objects. (LaTeX formatting was already implemented.)

2015-08-29:
   - Fix CSS class of ftable formatted in HTML
   - Implement `ftable_matrix` objects - objects composed from several `ftable`s.

2015-08-26:
  - Added `percentages` function to easily create tables of percentages.

2015-08-18:
  - Fix bug that presented Stata 9-10 files to be successfully imported.

2015-08-16:
  - Fix that allows to import ANES 2008 data with the provided buggy SPSS script files.

2015-08-03:
  - Allow specification of `show.baselevel=`, and `baselevel.sep=` as arguments of `mtable()`

2015-07-21:
  - Make `write.mtable` more coherent.

2015-07-20:
  - Eliminate buglino that causes knitr to gobble up stars (not those in the sky, 
    but those used to indicate levels of statistical significance)

2015-07-18:
  - New `signif.symbols` argument to `mtable`, that allows to change
    the "level of significance symbols"
  - Implemented `[` operator for `mtable` results as well as
    the `c` (concatenation) operator.

2015-07-15:
  - New `gs.options=` argument to `mtable`, that allows to pass optional
    arguments to getSummary.

2015-07-14:
  - Added vignette about survey items (objects of class "item")

2015-07-13:
  - Add basic documentation for 'summaryTemplate' generic function.
  - Add documentation for 'dsView'.
  - Fix codoc mismatches pointed out by R CMD check.
  - 'spss.fixed.file' now checks whether there are undefined variables in 'varlab.file' etc.

2015-07-10:
  - Correct `sourcename` default argument of `collect` methods.
  - Get rid of duplicate aliases in documentation.
  - FIX: `coef.tab` not `coef.table`
  - Add documentation for `datetime.item` objects (merged from an earlier development branch)
  - Add documentation for data set manipulation (`merge` etc.)
  - Make `cases` more fussy about logical conditions.

2015-07-02:
  - Provide HTML formatting of data frames.

2015-06-15:
  - New option 'show.baselevel' for function 'mtable' that allows to decised whether the baseline category of 
    dummy variable shows up in the table.

2015-06-13:
  - Scalars created by 'within' expression are now retained constant items in data.sets
  - 'sourcename' argument of collect methods for data frames and data.sets now has a 
    coherent default.

2015-06-08:
  - 'mtable_format_html' works again
  - Slight reorganisation of 'mtable' and associated formatting functions.
  - Nicer grouping of coefficients of multiple-equation models
  - 'applyTemplate' now returns empty strings for undefined quantities.

2015-06-02:
  - Moved 'toLatex' method for 'data.set' objects into different file. Added 'toLatex' method for data frames.

2015-06-01:
  - Completely refactored 'mtable' - more flexible handling of models with varying numbers of equations; 
    some formatting is delayed and moved to the 'mtable_format' functions.

2015-05-31:
  - Added toLatex-method for data.set objects.

2015-05-28:
  - Added missing 'drop=FALSE' in format_html.codebookEntry
  - 'codebookEntry' can now deal with factors that have no non-missing levels

2015-05-16:
  - 2015-05-16: Added support for date-time "item" objects (from former 'codebooks' package)
  - dsView, cbind, rbind, collect, and merge methods for "data.set" objects copied from former 
    'codebooks' package to 'memisc'.

2015-05-14:
  - Fixed bugs introduced by rewrite of 'import-internal-spss-syntax.R'.
    Dependency 'MicroDatosEs' should work okay now.

2015-05-12:
  - Fixed a bug introduced by side-effects of manipulating the arguments of 'readfixed'. 
  - Make sure that codebook entries for items in external data sources are complete.
  - Add dots to 'format_html.codebookEntry' formal arguments.
  - Improved `toLatex` formatting of `ftable`s
  - Set version number to 0.98.1

2015-05-11: Make memisc:::trimws work without changing locale
  - Complete rewrite of major parts of 'import-internal-spss-syntax.R': 
    Handling of SPSS syntax files for fixed format data files is changed from a presumptious vectorising approach
    to an old-school but more flexible and stable Turing approach. 
  - 'spss.fixed.file()' should work for a wider range of situations.

2015-05-10:
  - Use endash in HTML format of codebooks, correct removal extra styling of variable names

2015-05-09:
  - Remove hardcoded styling of variable names in codebook entries.
  - Codebook support for numeric vectors and factors
  - Fix console output of data.set objects

2015-04-29:
  - Step up version number and date.
  - Removed functions panel.errbars, Termplot, Simulate, and associated stuff (buckets etc.)

2015-04-28:
  - Documented 'format_html' method for codebooks
  - Added HTML formatting of codebooks.
  - Bugfix in 'codebook-methods.R': specify sensible defauls for 'toprule' and 'midrule' in format,codebookEntry-method.
  - Added dumentation for 'format_html' methods for "ftable" and "mtable" objects.
  - Added documentation for 'format_html' function
  - Documented 'wild.codes' function
  - Fixed as.data.set,list-method, added documentation for as.data.set,list-method

2015-04-26:
  - Changed vignette engine to 'rmarkdown' as this gives better output.
  - Added html vignette 'ANES 1948'
  - Fix html formatting of mtables
  - Bugfix of format_html.ftable: Now digits after the dot are handled correctly.
  - Some adjustments of HTML formatting of "mtables"
  - Support for HTML formatting of "ftables"
  - Added function 'wild.codes' to find, well, wild codes.

2015-04-20:
  - Fix collect method of "data.set" objects, by providing "as.data.set" method for lists.

2015-04-09:
  - Improvement of header formatting of mtables for LaTeX

2015-04-07:
  - Shorter vignette for html output of mtable results.
  - Update README.md
  - Create README.md
  - More modular code for HTML formatting
  - Added knitr vignette for HTML output example 
  - Corrected NEWS.Rd
  - Minor correction of format of ANES48 vignette

2015-04-06:
  - Improved HTML output of 'mtable' results  - alignment on decimal signs and straight borders. 
  - Updated documentation of 'mtable' results.

2015-04-05:
  - New coefficient style - confidence intervals right of estimates
  - Some cleanup in mtable; special formatting functions for mtable objects (needs testing)

2015-03-07:
  - Renamed 'getSummary.expCoef' into 'getSummary_expCoef' to avoid R CMD check fussiness.

2015-02-14:
  - Modified file structure somewhat and added some lines do clarify credit for code.
  - Added documentation for 'as.character' method for 'codebook' objects.
  - Implemented 'Write()' convenience function for 'descriptions' objects.

2015-02-13:
  - Removed 'aggregate.formula' method as it conflicts with a method of the name that was included
    into the 'base' R package more or less recenlty.
  - Removed 'include' and 'uninclude' since it modifies (in a benign way though) the global namespace,
    which is discouraged by recently changed CRAN policies.

2015-01-26:
  - Added 'as.character' method for 'codebook' objects, to allow conveniently write the to files using 'writeLines'

2015-01-25:
  - Explicitly export S3 method as required by package check by more recent R version
  - Added github URL
  - Listed packages that are 'enhanced' by supporting them for `mtable`

2014-12-14:
  - Correct buglet in 'recode' method for numeric vectors

2014-11-04:
  - Fixed silly bug introduced in recent update: Long variable names in SPSS "sav" files should work again now.

2014-10-11:
  - added method for "merMod" objects to the getSummary() generic function. mtable() now 
    should be able (again) to handle estimation results produced by lmer() and glmer() from 
    package 'lme4'.
  - recode() handles character vectors in a more convenient way: They are converted
    into factors with sorted unique values (after recoding) as levels.

2014-09-08:
  - 'getSummary.glm' now takes care of missing values in weights of glm objects.

2014-08-19:
  - modified some code that Prof Ripley thinks looks 'suspicious'.

2014-06-24:
  - Change of SPSS variable names to lower case is now optional (but still the default).

2014-03-10:
  - Removed modification to getSummary.clm that slipped in from a development branch.

2013-12-06:
  - fixes out-of-bound error and memory leak in 'pspp-portableStream-for-R.c'
    reported by Prof. Ripley.  Eliminated calls to non-exported functions from 'lattice'.
    Marked all example code of 'Simulate' as 'dontrun' to save time during package checks.
    Also did some cosmetic changes to the package vignette 'anes48'.

2013-09-19:
  - fixes of some namespace import issues.

2013-07-28:
  - fix of 'data.set()' and 'Table()' functions.
    Thanks to Gabor Grothendiek and Jeremy Jackson 

2013-03-20:
  - getSummary method for "mer" objects was broken and now is fixed.

2013-02-10:
  - getSummary methods for "ivreg", "tobit" (from package "AER"),
    "hurdle", "zeroinfl" (from package "pscl"),
    "betareg" (from package "betareg"),
    "multinom" (from package "nnet"),
    contributed by Achim Zeileis,
    were included into the package, but had not been exported yet
    from the package namespace.
    This is now fixed.

2013-01-27:
  - Minor improvement of 'codebook' output for external SPSS data sets.
  - Fixed a bug in 'dumbswap.c' discovered while building a windows binary.

2013-01-26:
  - Included getSummary methods
     - for objects of classes "ivreg", "tobit" (from package "AER"),
                              "hurdle", "zeroinfl" (from package "pscl"),
                              "betareg" (from package "betareg"),
                              "multinom" (from package "nnet"),
                              contributed by Achim Zeileis
    - for objects of classes "polr" (from package "MASS"),
                             "clm" (from package "ordinal"),
                             "simex" (from package "simex"),
                             contributed by Christopher N. Lawrence
    - for objects of classes "coxph", "survreg" (from package "survival"),
                             "aftreg", "phreg", "weibreg" (from package "eha"),
                             contributed by Jason W. Morgan,
    - for exponentiated coefficients in glm-models,
                             contributed by Dave Atkins.
  - Made the 'spss.fixed.file' importer function more robust:
      - It works now, even if 'varlab.file', 'codes.file',
        or 'missval.file' arguments are missing. (This was broken by an earlier revision.)
      - It now handles value labels of character string variables correctly.

2013-01-04:
  - Updated contact email address. Corrected "initialize" method for "data.set" objects.

2012-02-14:
  - Replaced calls to .Internal(sample()) by calls to sample.int() and eliminated
    call to .Internal(int.unzip()). The package now depends on R >= 2.9.0

2012-01-25:
  - Fixed reading of value labels from SPSS system files - no garbage at the end of the labels.

2011-12-07:
  - Fixed "getSummary" and "getSummary.expcoef" S3-generic functions.

2011-11-06:
  - Added a 'rep' method for 'item.vector' objects.
  - Applied patch sent by Masahiro Hayashi <rinm884@gmail.com> to fix bugs
    in 'vector' method of 'recode'.

2011-10-05:
  - Adapted the package to R 2.14.0 (by eliminating calls to 'require'
    in the '.onLoad' function)
  - Changed the methods for data frames, 'data.set' and 'importer' objects
    of the function 'sort' from S3 to S4

2011-08-26:
  - Fixed a bug in internal function 'compress.storage.mode.double' that caused
    it to create NA's where it should not.
  - Fixed a bug in the print method for items that caused it to report their length
    incorrectly

2011-08-15:
  - Made internal functions of 'spss.fixed.file' robuster against syntax errors
    in the spss-controls files (codes.file, labels.file, etc.)

2011-06-17:
  - Removed mkHelpDir() that has not worked for a while because of architectural
    changes in R.
  - Removed some unused variables from the C-code that caused warnings.
  - Corrected a bug in format.mtable() misusing the function character()
    and some other buglets reported by NOTE's in automatic runs by CRAN.

2011-05-29:
  - Applied patch sent by Christopher N. Lawrence, Ph.D. <c.n.lawrence@gmail.com>:
    Added support to mtable() for clm and clmm ordinal models in the ordinal package.
    Updates getSummary to support mer objects as generated by recent versions of the lme4 package.
    Adds the Aldrich & Nelson pseudo-R2 (Aldrich and Nelson 1984)
    to mtable output for various categorical models.

2011-05-14:

  - Fixed problems with SPSS system files that do not contain an 'info_flt64' record. This problem led to zeroes in the data section of the file being erroneously interpreted as *sysmis*.
  - Fixed 'show' and 'print' methods for 'data.set' objects - they no do fit the width set by options(width=...), without ugly line breaks.
  - Changed the default value of the 'compress.storage.modes' option for the
    subset and bracket methods o 'importer' objects to FALSE, which is much
    more safer.

2010-07-16:
  - fixed a bug in 'Simulate' that caused it to fail in certain situations
    in R version >= 2.11.

2010-04-12:
  - 'toLatex(mtable(<something>),useDcolumn=FALSE)' should now produce useable
    LaTeX output ($'s are added to the table entries).

2010-04-09:
  - 'By' works now with 'genTable' and 'aggregate.formula'. It gains a 'data.frame'
    method, which better cooperates with calls to 'lm'-like functions. Their results
    get a 'data' component containing the subset of the data frame that was given
    as 'data' argument to the call of 'By'.

2010-04-08:
  - fixed bug in 'cases'. It works now with target expressions even with 'within'.
  - 'write.mtable' now handles a 'forLaTeX=TRUE' option in a more convenient way.
  - 'mtable' now handles objects of classes "polr" and "simex", thanks to a contribution
    from Christopher N. Lawrence (Texas A&M International University)
  - support for 'mtable' output for "glm" objects with exponentiated coefficients
    and prelininary support for "lmer" object added, thanks to a contribution from
    Dave Atkins (Unversity of Washington). Also the coef.styles are changed to be
    more convenient.

2010-04-07:
  - fixed bug in C-level function 'read_sysfile_subset' that caused spurious errors
    while importing SPSS 'system' files with long strings.

2010-03-18:
  - patch to trimws to make it locale-neutral

2010-03-11:
  - fixed bugs in mtable and format.mtable. Short 'summary.stats'
    arguments or 'summary.stats=FALSE' should work now.

2010-03-06:
  - fixed bug in genTable/percentage: total counts disappeared if
    a factor level had the same name as the name given to the total count
  - changed the default value of the 'digits' option of the matrix method
    of the toLatex generic to the more sensible getOption("digits")
  - the matrix method of the toLatex generic gains new optional arguments
    'show.vars', 'show.xvar', and 'show.yvar', which determine whether
    the name of the row and/or the column variable should be shown if
    the matrix is a 2-dimensional table.

2010-03-02:
  - fixed minor bug in 'cases' function

2009-11-11:
  - correction bug in .unLoad function.

2009-10-14:
  - removed 'extended' arguments from calls to 'regexpr' and 'sub' etc. (for compatibility
    with R 2.11)

2009-08-19:
  - corrected documentation for mtable (\usepackage instead of usepackage)
  - completed documentation for contrasts generic
  - fixed bug in the internals for 'spss.portable.file':
    'SPSS Por' to ascii and 'SPSS Por' to integer translation tables
    were one item too short.

2009-06-02:
  - as.factor methods for item vectors adapted to new factor function
    in R 2.10.0
  - Fixed bug in sample.data.frame and sample.data.set methods:
    Samples were drawn only from the first m observations, where m is
    the number of variables in the data frame/data set.
