**********************************************************************************
**********************************************************************************
** 
**  WISHLIST: 2008-01-13 for xts
**
**********************************************************************************
**********************************************************************************

* Write more test cases to better 'test' the functions, specifically tests
  to check on xts class persistence


* time<- method for xts - alias for index<-


* index<- method to check for time-date requirement


* [.xts<- replacement method


* cbind/rbind xts methods required to maintain xts object


* aggregate.xts (dispatch to NextMethod or call aggregate.zoo and
  re 'xts' results

 
* diff.xts method  [not needed: jar]


* as.data.frame.xts, as.matrix.xts
  as.zoo.xts, as.its.xts,
  as.timeSeries.xts, as.irts.xts

  above should handle clean coercion to the requested classes,
  so that it would be possible to do the following:
    x # a timeSeries object
    identical(as.timeSeries(as.xts(x)),x)

     -- or --

    as.ts(as.xts(x))

  possibly rely on zoo methods instead, though may be able to do better in
  certain cases. timeSeries may be the tough one - as it has unique attributes
  automatically generated, that *should* just be copied.

  'as.ts' is a bad idea to rewrite.  The current method works great
  (as.ts.default or as.ts.zoo???)

  The most compelling reason is to allow for temporary conversion
  to class type required by particular functions - esp. functions
  in PerformanceAnalytics and the Rmetrics bundle, though a reliable
  coversion to 'ts' for many base functions would be also useful

  essentially a mechanism to allow for the same conversion as.* is
  used for elsewhere.

* indexClass and indexClass<- methods allow for changing index class,
  at present only simple conversions work, as no additional args
  are passed to as.POSIXct,etc methods.  Need to fix this.

  Additionally indexing to different formats needs to be done:

      as.yearmon.xts, as.yearqtr.xts

* per suggestions from Brian Peterson automatically
  handle mon/qtr as yearmon and yearqtr 
  
* look into chron, dates (& yearmon, yearqtr)
  -- this can't be extensible, as we couldn't test for time nature...
  -- but we *should* cover all *known* date handling in R at any given time

* create a matrix of time-series types and attributes for comparison
  documentation

* write an introductory vignette for 'xts', possibly an article for R news

* web page on R-Forge

* a generic extension function ?extend.xts?
