* stock splits
  [ idempotent backpopulation, which is done in 0.5.0, eases this a little ] 

* annualised returns
  [was done in status command, currently turned off; probably needs another
   switch to allow a date after which we annualise, say a year]
  [turned back on again]

* clean up code

* more error checking on calls to DBI et al

* read more info from the rc file ~/.beancounterrc such as host, 
  port, user, ... 
  [ mostly done; this grows organically on an as-needed basis ]

* allow "restrictions" on beancounter operations, eg do a portfolio update at
  13:00 EST with the restriction of doing it only for shares with currencies
  that are GBP or EUR (or SEK or DKK ...) to get the European market
  [ done in 0.2.0, also added 'extrafx' argument ]

* there is buglet that rears its head when a stock is added to portfolio 
  during business hours as its current price is also added for that day ...
  which creates a conflict when that stock is later updated during the normal
  update. This is difficult to circumvent as we don't want to reinvent
  cron within beancounter. Either we do update for the current day (as we do
  now, which has the problem) or don't (in which case we might never get that
  datum if we updated after the daily cron run)
  [ cf below on idempotent operations, should no longer be an issue ]

* fx quotes are currently retrieved via Yahoo from the Philadelphia exchange
  -- but that means no quotes are available on US holidays. 

* We also have no access to historical FX data for backpopulation
  [ as of 0.6.0 we do, albeit at a grizzly two-decimals precision ]

* there is also a plain error somewhere as eg a DEM based portfolio sees a
  return even though a EUR traded stock didn't move -- can't really happen
  as DEM/EUR rates are fixed. Maybe add an crosscurrency table/hash and do
  no ask Yahoo! if cross-rate known 
  [ not an issue if portfolio is reported in EUR ]

* Yahoo seems to return bad date information every now and then. A
  "scrubbing" between retrieval and databasing might help: we can check if
  other data points for the same exchange have a different date and then use
  this as an override. Might be overkill.
  [ partly done; now have --forcedate option ]

* Allow for cash holdings
  [ partly done; needs support for adding/changing/deleting positions ]

* Options (.X) volume adjustment (ie an option contract is over 100 shares)

* MySQL support
  [ added in 0.4.0 following Mathias' patch with the initial MySQL schema ]

* Beancounterrc manual page
  [ there is an updated section in the beancounter(1) manual page ]

* Idempotent operation, ie call update several times a day
  [ added in 0.4.0 based on Mathias' patch ]

* Versioning of db schema, ie add new table "version" with one column
  and maybe a date

* update_beancounter needs to reflect the post-0.3.1 table changes
  [ done, tested against 0.1.1, 0.3.0 and 0.3.1 (with PostgreSQL) ]

* beancounter regression test data set, data "feeding" program and test 
  routines
  [ Mathias is looking into this ]

* updates for stocks in stockinfo but not in portfolio; maybe via an option
  to update (eg --fullupdate)
  [ done in 0.4.0 via rewriting of the symbol selection SQL query ]

* new command for partial deletes (ie from portfolio, but leave stockinfo and
  stockprices). Or maybe make partial delete the default, and only nuke from
  stockinfo/stockprices if --purge option given

* GUI:
  -- one choice would be via perl/Tk [ but IMHO Tk's canvas widget is too
     primitive ]
  -- another via R's (www.r-project.org) tcltk package; R's advanced graphics
     and analysis functions could shine there [ have prototypes ]
  -- web-based via mod_perl [ Mathias' suggestion ]

* (really distant wishlist feature) build portfolio up from stock trades, 
  have analysis driven from transactions

* new columns "soldprice","solddate" to compute returns on completed 
  transactions
 
* suppress noisy error messages when we have not enough data for the risk 
  report

* added 'active' boolean field to stockinfo
  [ done in 0.5.0 ]

* restrictions don't aggregate right if they only "partly" hit a holding
  (eg holding a stock in two accounts, blocking one --> still get total)