[GNC-dev] New balsheet (and P&L report), API considerations, and (slow?) KVP in Account.cpp

[Christopher Lam]

I think I'll forego a noclosing_balance upgrade in C.

According to https://bugzilla.gnome.org/show_bug.cgi?id=570042 the 
"Closing Entries" transactions did not always receive a special 
KVP-based flag until the commit on 
https://github.com/Gnucash/gnucash/commit/4b2800145 on datafiles 
generated from 27-11-2008 and later.

So, to find and exclude these closing entries, we can use the following 
strategies:

 1. xaccTransGetIsClosingTxn, will query the KVP flag
 2. Run a QofQuery and add xaccQueryAddClosingTransMatch (also queries
    the KVP flag)
 3. Free-text search "Closing Entries" to seek these old closing entries

So a P&L report will necessarily need to do both filtering for KVP and 
free-text search.

The multicolumn report will be different enough from the old 
balance-sheet and income-statement that I think I should spin it off 
into a different report altogether, and the others will be sunsetted. 
Hopefully this new report will be broadly acceptable because the old 
reports have a *lot* of supporting unintelligible old code, especially 
to handle closing-entries as above.

C

On 24/06/18 00:51, Christopher Lam wrote:
> Hi John, the split->noclosing_balance is updated in 
> xaccAccountRecomputeBalance. Will continue copypasta coding until it 
> works!
>
> On Sat, 23 Jun 2018, 23:56 John Ralls <jralls at ceridwen.fremont.ca.us 
> <mailto:jralls at ceridwen.fremont.ca.us>> wrote:
>
>
>
>     > On Jun 22, 2018, at 9:42 PM, Christopher Lam
>     <christopher.lck at gmail.com <mailto:christopher.lck at gmail.com>> wrote:
>     >
>     > Hi All
>     >
>     > I'm working through balance-sheet.scm and overhauling this report.
>     >
>     > At the same time, I can see that balance-sheet.scm and
>     income-statement.scm can be merged together.
>     >
>     > After all
>     >
>     > * balance sheet = asset/liability/equity balance at date X,
>     > * income statement = difference in income/expense balances at
>     date X and Y
>     >
>     > The issue I have is that the balance sheet can call
>     xaccAccountGetBalanceAsOfDate(acc,date) directly to nicely
>     retrieve the balance, whereas income statement cannot directly
>     call it because it also includes closing transactions.
>     >
>     > My proposal is to augment xaccAccountGetBalanceAsOfDate to
>     accept a 3rd boolean argument i.e.
>     xaccAccountGetBalanceAsOfDate(acc,date,ignoreclosing) which will
>     selectively produce the balance, ignore closing transactions.
>     >
>     > This is partly done in the last commit of
>     https://github.com/christopherlam/gnucash/tree/maint-create-API-call-for-noclosing-balances
>     - it will augment API everywhere, and also modify Account.cpp,
>     especially xaccAccountRecomputeBalance which will call
>     xaccTransGetIsClosingTxn(xaccSplitGetParent(split)) for each split
>     in the account to determine closing status and cache the balances
>     for each split. See draft on
>     https://github.com/christopherlam/gnucash/tree/maint-create-API-call-for-noclosing-balances
>     - this currently fails because I'm not good at C.
>     >
>     > *My query is whether it will be too expensive to call
>     xaccTransGetIsClosingTxn for each split in Account.cpp to produce
>     the split->noclosing_balance, which will be crucial to calculate
>     income between two dates.*
>     >
>     > *Currently this 'closing-txn' filter is done in
>     income-statement.scm via a "Closing Entries" string/regex filter
>     and xaccTransGetIsClosingTxn calls, which IMHO is not ideal either.
>     > *
>     >
>     > Any help welcome!
>     >
>     > P.S. if this last commit is removed, the
>     https://github.com/christopherlam/gnucash/tree/maint-create-API-call-for-noclosing-balances
>     branch contains work-so-far for updated balance-sheet. Screenshot
>     https://screenshots.firefox.com/eelmGQrGCtcP6bqC/null - see the
>     multilevel subtotals were previously horizonal, are now vertical;
>     original-currency amounts, and multiple-date balance sheets.
>     >
>     > _
>
>     Chris,
>
>     Only profiling will tell how much of a performance hit this creates.
>
>     I don’t see where in your commit you’re computing the split’s
>     no-closing balance. Perhaps that’s why it doesn’t work?
>
>     Regards,
>     John Ralls
>