previously fmtnumeric would convert gnc_numeric to double. simplifying
the code had omitted this conversion, and exact fractions were
rendered. ensure decimal is rendered.
This is a *subtle* bug.
1. dates-list is the list of report-dates at periodic intervals
2. if an interval report (i.e. inc/exp) is chosen, we want to omit the
last one, because the report analyses changes in value between
adjacent dates.
3. dates-list was appropriately redefined to be 1 element shorter than
the original dates-list.
4. date-string-list (for display labels) was meant to pick up the
shorter dates-list, formatted via qof-print-date.
Unfortunately the let statement had defined both dates-list and
date-string-list at the same time, therefore date-string-list did not
pick up the redefined dates-list. Using let* instead means it uses the
shorter dates-list, and omits the last element as intended for display.
the flags warn-no-price and warn-price-dirty are report-specific
warnings to flag price difficulties. these warnings must now be
recorded in table-add-stock-rows because they are not accessible
globally.
Combine the best of both approaches and allow jumping to invoice if
either
- the current txn is an invoice transaction
- the current split has an invoice associated with it
Those are the only two conditions that allow us to uniquely
identify an invoice from a register.
A future extension could be to allow a jump if any split
of the current txn has an invoice associated with it.
This would need user interaction if more than one split
is linked to an invoice. A payment transaction for example
can pay more than one invoice and hence can have more than one
invoice related split.
Column widths of 0 are not saved by the table and so revert back to
calculated widths when reloaded. This change sets the column width to
1 when the right border is dragged to the left border to hide columns.
By doing this, a visual indication of a hidden column is seen as the
column border is now wider by 1px.
The single-character column heading "C" is used for "Color" in one
place,
and for "Cleared" in another place. Obviously this must be fixed by
adding
context, otherwise grossly wrong translations will show up (as is
currently
the case for German and the "Cleared" column in the import matcher)
PR #744
