Changed in xts 0.12-0:

o  All the index-attributes have been removed from the xts object and are now
   only attached to the index itself (#245). We took great care to maintain
   backward compatibility, and throw warnings when deprecated functions are
   called and when index-attributes are found on the xts object. But there
   still may be some breaking changes lurking in edge cases.

   o @SamoPP found one edge case (#297) where an error was thrown when index()
     was called on an xts object with an index that had no tclass attribute.
   o ...which led Joshua to find that the index setting functions did not
     always copy index attributes (#305).

o  Several binary operations (e.g. +, -, !=, <, etc.) on variations of
   uncommon xts objects with other xts, matrix, or vector objects, could
   result in malformed xts objects (#295). Some examples of the types of
   uncommon xts objects: no dim attribute, zero-width, zero-length.

o  Calling as.matrix() on an xts object without a dim attribute no longer
   throws an error (#294).

o  merge.xts() now honors check.names = FALSE (#293).

o  The possible values for major.ticks, minor.ticks, and grid.ticks.on in the
   Details section of ?plot.xts have been corrected. Thanks to Harvey Smith
   (@harvey131) for the report and patch (#291).

o  as.zoo.xts() is now only registered for zoo versions prior to 1.8-5. Methods
   to convert an object to another class should reside in the package that
   implements the target class. Thanks to Kurt Hornik for the report (#287).

o  .parseISO8601() no longer has a potential length-1 logical error. Thanks to
   Kurt Hornik for the report (#280).

o  endpoints() now honors k > 0 when on = "quarters". Thanks to @alkment for
   the report (#279).

o  Performance for the period.XYZ() functions (sum, prod, min, max) is much
   faster (#278). Thanks to Harvey Smith (@harvey131) for the report and
   examples.

o  merge.xts() now creates shorter column names when passed unnamed objects.
   This is now consistent with zoo (#248).

o  Time-of-day performance is ~200x faster, thanks to StackOverflow
   user3226167 (#193).

Changed in xts 0.11-2:

o  The to.period() family of functions now use the index timezone when
   converting intraday index values to daily values (or lower frequency).
   Thanks to Garrett See and Gabor Grothendieck for the reports (#53, #277).

o  Make column names for merge() results with unnamed objects shorter and more
   like zoo (#248). This also makes na.fill() much faster (#259).
   BREAKING: This may break existing code for integer unnamed objects.

o  Fix subset when 'index(x)' and 'i' contain duplicates. Thanks to Stack
   Overflow user 'scs' (https://stackoverflow.com/users/4024268/scs) for the
   report, and Philippe Verspeelt for debugging (#275).

o  Fix if-statement in xts constructor that may use a logical with length > 1.
   Thanks to @HughParsonage for the report and PR (#270, #272).

o  Register shift.time.xts() method. Thanks to Philippe Verspeelt for the
   report and PR (#268, #273).

o  Conditionally register S3 methods for as.timeSeries.xts() and as.fts.xts()
   when their respective packages are available (as requested by CRAN). Note
   that this means these two functions are no longer exported. This may break
   some existing code that calls the methods directly, though 'best practice'
   is to let method dispatch determine which method to invoke.

Changed in xts 0.11-1:

o  Fix regression in .xts() that caused 'tclass' argument/attribute to be
   incorrectly set to POSIXct regardless of user-provided value. Thanks to
   @Eluvias for the report and Tom Andrews for the PR (#249, #250).

o  Fix performance regression when xts object is subset by a date-time vector.
   Thanks to Tom Andrews for the report, and the PR to fix the bug in my patch
   (#251, #263, #264).

o  Restore behavior from 0.10-2 so subsetting an empty xts object by a date-
   time vector returns an empty xts object instead of throwing an error.
   Thanks to @alkment for the report (#252).

o  Add ability for merge.xts() to handle multiple character or complex xts
   objects. Thanks to Ken Williams for the report (#44).

o  Add ability to use "quarters" to specify tick/grid mark locations on plots.
   This ran but produced an incorrect result in 0.10-2 and threw an error in
   0.11-0. Thanks to Marc Weibel for the report (#256).

o  Fix illegal read reported by valgrind. Thanks to Tom Andrews for the report
   and PR (#236, #264).

Changed in xts 0.11-0:

o  Fix make.index.unique() to always return a unique and sorted index. Thanks
   to Chris Katsulis for the report and example (#241).

o  Add window.xts() method and completely refactor the internal binary search
   function it depends on. Thanks to Corwin Joy for the PR, tests, review, and
   patience (#100, #240).

o  Better axis tick mark locations for plots. Thanks to Dirk Eddelbuettel for
   the report (#246). Also incorporate axTicksByTime2() into axTicksByTime() to
   reduce code duplication from the migration of quantmod::chart_Series() to
   xts::plot.xts() (#74).

o  Add details to plot.xts() parameters that are periodicity, now that RStudio
   has argument completion. Thanks to Evelyn Mitchell for the PR (#154).

o  periodicity() now warns instead of errors if the xts object contains less
   than 2 observations (#230).

o  first() and last() now keep dims when they would otherwise be dropped by a
   regular row subset. This is consistent with head() and tail(). Thanks to
   Davis Vaughan for the report (#226).

o  Fix subset when ISO8601 string is outside the valid range, so it returns no
   data instead of all rows (#96).

o  Avoid partial name matches from parse.side() (inside .parseISO8601())
   results that are passed to firstof() and lastof(). Thanks to @gp2x for the
   report and the patch (#231).

o  na.locf.xts() now loops over columns of multivariate objects in C code,
   instead of in R. This should improve speed and memory performance. Thanks to
   Chris Katsulis and Tom Andrews for their reports and patches (#232, #233,
   #234, #235, #237).

o  Change plot.xts() default 'pch = 0' (rectangles) to 'pch = 1' (circles) so
   it looks more like base and zoo plots (#203).

Changed in xts 0.10-2:

o  na.locf.xts() and na.omit.xts() now support character xts objects. Thanks to
   Ken Williams and Samo Pahor for the reports (#42).

o  na.locf.xts() now honors 'x' and 'xout' arguments by dispatching to the next
   method (#215). Thanks to Morten Grum for the report.

o  coredata.xts() now functions the same as coredata.zoo() on zero-length
   objects, and only removes xts-related attributes (#223). Thanks to Vincent
   Guyader for the report.

o  plot.xts() no longer ignores 'col.up' and 'col.dn' when 'type="h"' (#224).
   Thanks to Charlie Friedemann for the report. This was inadvertently broken
   as part of the fix for #210.

Changed in xts 0.10-1:

o  'ylim' values passed to 'addSeries' and 'addPolygon' via '...' are now
   captured and honored (#220).

o  'addPolygon' now checks for ylim of zeros, as 'addSeries' does (#164).

o  The 'base::as.Date.numeric' method is no longer over-ridden. The exported,
   but not registered, method in zoo should prevent any change in behavior.

o  Series added to an existing plot are now given the same index values as
   the main panel (#216). There still may be some weird behavior if the new
   data does not have observations within the timespan of the main panel data,
   but no observations on the same timestamps.

o  Existing 'par' values are now captured and reset before returning from
   plotting functions (#217).

o  User-defined 'col' values are now honored when 'type="h"' (#210).

o  Values passed to plotting functions are now copied from the calling
   environment. This enables plotting arguments to be objects passed
   through multiple layers of function calls.

o  indexFormat is now generic, consistent with indexFormat<- (#188).

o  Calling as.matrix() on a zero-width xts object now behaves consistently
   with zoo, and no longer throws an error (#130).

o  Fix weird result in merge.xts() when 'fill' argument is NULL or a zero-
   length vector (#261).

o  Fixed bug in endpoints() due to sub-second representation error via using
   integer division (%/%) with non- integer arguments (#202).

o  endpoints() gained sub-second accuracy on Windows (#202).

o  plot.xts() no longer errors when called on an object containing a constant
   value. It chooses ylim values +/-20% from the series value (#156).

o  plot.xts() now places y-axis labels in the same location on the plot,
   regardless of data periodicity (#85).

o  rbind.xts() now throws an error if passed an xts object with different
   number of observations in the index and data (e.g., zero-width) (#98).

Changed in xts 0.10-0:

Major changes include:
o  A new plot.xts() that is incompatible with earlier versions of plot.xts().
o  Moved development from R-Forge to GitHub.
o  New xts FAQ.

Other, less disruptive changes include:

o  merge.xts() now throws an error if the index contains non-finite values
   (#174).

o  Constructors xts() and .xts() now ensure order.by and index arguments do not
   contain non-finite values. Many xts functions, most notably merge.xts(),
   expect all index values to be finite. Missing index values usually indicate
   an error, and always occurred at the end of the index (#173, #194, #199).

o  Fixed bug in endpoints() when called on sparse data that have the same month
   and day, but different years (#169).

o  Fixed bug in [.xts did not do the same checks on logical xts objects as it
   does for all other data types (#163).

o  Fixed bug that caused split.xts() to error if 'f' is a character vector with
   more than 1 element (#134).

o  Fixed bug that crashed R if 'k' argument to lag.xts() was not an integer and
   would be NA when coerced to integer (#152).

o  period.apply() now checks to ensure the object's index is unique and sorted,
   and sets INDEX <- sort(unique(INDEX)) if it is not. It also ensures INDEX
   starts with 0 and ends with NROW(x) (#171).

o  All references to the 'its' package have been removed, since it is now
   archived on CRAN at the request of its maintainer.

o  Fixed bug that crashed R when merge.xts() was called on an empty xts object
   and more than one non-xts object (#157).

o  Fixed bug that did not set the index's tzone attribute to UTC when
   index<-.xts or indexClass<- were called and 'value' did not have a tzone
   attribute (#148).

o  Fixed a bug in endpoints() that caused incorrect results if the index was
   less than the epoch (#144).

o  Fixed a bug that caused diff.xts() on a logical xts object to return an
   object with a POSIXct index.

o  index.xts() works even if the package containing the class for the index
   is not attached (it needs to be loaded, however).

o  [.xts now returns NA if a one-column xts object is subsect by NA, instead
   of throwing an uninformative error (#97).

o  Fixed bugs that would crash R when [.xts was called a certain way and 'j'
   contained NA values (#97, #181).

o  Fixed a bug in endpoints() where 1 second would be subtracted for any date
   in the year 1969. The subtraction is only supposed to occur on
   1969-12-31 23:59:59.9... to work around behavior in strptime().

o  timeBasedSeq() now honors hour/min/sec 'BY' values (#91).

o  [.xts now throws an error if 'j' is character and not one of the column
   names. This is consistent with [.zoo and [.matrix (#48).

o  timeBasedSeq() now works correctly when resolution is "days" the sequence
   includes a daylight saving time change (#67).

o  Explicitly set indexTZ="UTC" for all index classes that do not have a TZ
   (#66).  indexTZ="GMT" is also allowed.

o  Fixed as.xts() when called on an 'mts' object (#64).

o  Moved development from R-Forge to GitHub.

o  Fixed bug in to.period() that errored when name=NULL (#5937).

o  Fixed bug in .index* functions that did not account for timezones (#5891).

o  Fixed bug that allowed index<-.xts to produce an unsorted index (#5893).

o  Fixed bug so subsetting a zero-width xts object with a zero-length 'i'
   vector no longer returns an object with column names (#5885).

o  Updated [.xts to handle 'i' containing multiple zeros (e.g. subsetting by a
   "logical" column of an integer xts object).

o  endpoints() now errors if k < 0.

Changed in xts 0.9-7:

o  Fixed bug that caused logical operators on xts objects to drop the 'tzone'
   attribute (#2750).

o  Fixed bug that ignored 'which.i' argument to [.xts on zero-width xts
   objects (#2753).

o  Fixed bug where xts() does not sort 'order.by' if x is missing (#4775).

Changed in xts 0.9-6:

o  Fixed bug where setting dimnames to NULL would break as.xts() (#4794).

o  Added checks to period.sum/prod/min/max to ensure INDEX is in [0,nrow(x)].

o  Fixed missing argument to na_locf() in the C/C++ xtsAPI (Dirk Eddelbuettel).

Changed in xts 0.9-5:

o  Increased zoo dependency version to 1.7-10 for changes in C code.

o  Fixed several minor issues in the C/C++ xtsAPI (Dirk Eddelbuettel).

Changed in xts 0.9-4:

o  Fixed bug where the index was missing the 'tzone' attribute.

o  Fixed to.period() bug when 'indexAt' is "firstof" or "lastof". (bug #2691,
   patch #2710, thanks to Garrett See)

o  Fixed subsetting bug on zero-width xts objects that returned NA data and an
   NA index (#2669).

o  xts' merge() method now has 'drop' and 'check.names' arguments to match
   the zoo merge() method.

o  'index<-' now correctly handles UTC Date objects when resetting index
   values. '.index<-' behaved correctly.

o  xts' rollapply() method now handles the 'fill' argument.

o  Added several functions to the C/C++ API:
   - make_index_unique
   - make_unique
   - endpoints
   - do_merge_xts
   - na_omit_xts
   - na_locf

o  Fixed xts' rollapply() method when input has one column, but function
   output has more than one column.

Changed in xts 0.9-3:

o  No user-visible changes.

Changed in xts 0.9-2:

o  Added C/C++ xtsAPI (Dirk Eddelbuettel)

o  Added tzone() and tclass() functions as aliases to indexTZ() and
   indexClass(), respectively. Eventually will Deprecate/Defunct the former.

Changed in xts 0.9-1:

o  xts() now ignores timezone arguments when order.by is Date class, with a
   warning.

Changed in xts 0.8-8:

o  Modified str() output to make use of proper ISO-8601 range formating

o  Fixed bug in reclass() when 'tzone' of object is different than system TZ.

o  Fixed bug in xts() that dropped dims when 'x' is a 1-column matrix or
   data.frame.

o  [.xts no longer warns if 'i' is zero-length.

o  to.period() now checks that 'x' is not zero-length/width.

o  Fixed edge case in Ops.xts where two objects with no common index create
   an invalid 'xts' object with no index.

o  to.monthly() and to.quarterly() now default to drop.time=TRUE.

o  Internal .drop.time() now changes the index class to Date. This affects
   the to.period() family of functions.

o  Restore Date index/tclass conversion to POSIXct with a UTC timezone via
   integer division instead of double-precision division.

Changed in xts 0.8-6:

o  Revert Date index/tclass conversion to POSIXct with a UTC timezone to
   previous behavior as in 0.8-2.

Changed in xts 0.8-5:

o  A Date index/tclass is now internally converted to POSIXct with a UTC
   timezone ensure proper conversion regardless of user TZ settings.

o  tclass is now an argument to .xts()

o  Fix endpoints() properly handles millisecond time stamps (and microsecond
   on not Windows).

o  Subsetting zero-width xts objects now behaves like zoo, with NA values
   returned for valid row requests.

Changed in xts 0.8-2:

o  Fixed bug in lag() and diff() for character coredata.

o  Fixed subsetting bug that returned a contiguous chunk of data even when
   a non-contiguous 'i' was provided.

o  Fixed bug that ignored FinCenter/TZ for timeDate index

o  period.apply() now only sets colnames if the number of columns in the input
   and output are equal.

o  Fixed periodicity() when scale = "yearly"

o  Fixed [.xts when row-subsetting via a POSIXct vector, which returned an
   object full of NA.

o  Added '...' to axis() call inside of plot.xts() to allow for 'cex.axis'
   and 'cex.lab' to be passed in.

o  Fixed axes=FALSE issue in plot.xts().

o  Dependency now on 1.7-0 or better of zoo (R-forge at present)
   This build now links to C code moved from xts to zoo. At present
   this is only for zoo_lag (used in lag and lagts)

o  Added 'drop' and 'fromLast' arguments to make.index.unique().

o  Added adj.time() and shift.time()

o  Fixed na.locf() bug that would fill trailing NA larger than 'maxgap'
   observations (#1319)

o  Updated indexFormat() documentation and add an example

Changed in xts 0.8-0:

o  Fix print formatting (#1080)

o  Fix bug related to na.locf() and zero-width objects (#1079)

o  Add .RECLASS = FALSE after '...' for as.xts.*() methods. This makes all
   as.xts.*() methods one-way (i.e. not reclass-able). Objects converted to
   xts via try.xts() can still be converted back to their original class
   via relcass().

o  Fix bug that caused colnames to be dropped if object is subset by time
   that is not in the index.

Changes in xts 0.7-5:

o  try.xts and reclass now are more efficient on xts objects,
   no longer appending a .RECLASS attribute. This penalty 
   (copying) is shifted to classes that are internally converted
   to xts.

Changes in xts 0.7-4:

o  internal attributes of index are now maintaining
   timezone (tzone), time class (tclass) information.

o  `[.xts` method is now using new C code. This may revert
   back as character-based objects are not supported. Changed
   for future code refactoring into zoo, as well as performance
   gains on integer, double and logical values. Also added in
   checks for NAs.  drop=TRUE now works correctly in all known
   applications.

o  (cbind)merge.xts and rbind.xts now copy index attributes
   to handle internal changes to index characteristics (in C code)

o  indexTZ.Rd updated to provide information regarding internal
   changes.  Also indexTZ<- is now exported to facilitate 
   timezone changing of xts objects.

Changes in xts 0.7-1:

o  subsecond ISO8601 subsetting on dates
   before 1970 (epoch) is disabled. This is due to a bug
   in the R implementation of POSIX handling of fractional
   seconds pre-1970.  10 microsecond granularity is still
   functional for all other times. Thanks to Andreas Noack Jensen
   for the early bug report.

o  new 'tzone' arg in xts constructor and 'tz' in .parseISO8601
   allows for future support of non-system TZ dependent indexing

o  internal index attribute (numeric) now can have attributes
   set (tzone is currently the only one used in xts). These should
   remain during all xts operations. Still experimental.

o  naCheck has been exposed at the C level for use in packages
   "LinkingTo: xts".  See ?xtsAPI for more details.

Changes in xts 0.7-0:

o  A new NEWS file.  

o  print.xts now passes ...

o  endpoints speedup and bug fix (thanks Ismail Onur Filiz)  

o  na.omit bug on logical and NaN fixes (thanks Fabrizio Pollastri
   and Koert Kuipers)

o  fromLast=FALSE for na.locf.xts.  Matching to zoo. (thanks
   to Sandor Benczik)

o  LGLSXP support in leadingNA (R fun naCheck)

o  fixed logical and NA 'j' subsetting. Thanks Koert Kuipers.

o  as.xts and as.timeSeries fixes for timeSeries changes

o  merge and subset now support dimensionless xts (non-standard).
   merge segfault fixed when merging all 3 or more zero-width xts objects
   and only zero-width objects.  Thanks to Koert Kuipers for the report.

o  added which.i to [.xts to return i values found via
   ISO8601 subset string

o  new lines.xts and plot.xts, similar to methods in zoo

o  lastof now has sec to 10 microsecond precision, and subsec
   arg to append to secs.

o  xts() further consistency in NROW/index check

o  align.time error checks for positive n= values (thanks Brian Peterson)

o  toPeriod updates in C, almost exported. ~600-1200x faster

o  new lag_xts in C.  Increased speed and index flexibility.

o  endpoints 'days' bug fix 

o  .makeISO8601 function to create ISO8601 compliant string
   from xts objects
