forecast end date appears off by one

[simonmichael]

As I think we mentioned somewhere recently, when generating forecast transactions, hledger treats the end date as inclusive, rather than exclusive. It means that forecast transactions can end a day earlier than you would expect, and it's not mentioned in docs. Today I found it confusing, and I think we should make it consistent. Examples:

2021-07-26
  (assets)  1000

~ daily
    (assets)                                       1

reg/areg forecast reports end 2 days before the -e ENDDATE (or --forecast=..ENDDATE),
(rather than the usual 1 day before), unless there are also real transactions up to that date:

$ hledger-1.22 areg -w80 assets -b 2021-07-31 -e 2021-08-05 --forecast
Transactions in assets and subaccounts:
2021-07-31                      assets                           1          1005
2021-08-01                      assets                           1          1006
2021-08-02                      assets                           1          1007
2021-08-03                      assets                           1          1008

bal -D goes to 1 day before (since 1.21), but forecasted transactions still end 2 days before:

$ hledger-1.22 bal --transpose -D assets -b 2021-07-31 -e 2021-08-05 --forecast
Balance changes in 2021-07-31..2021-08-04:

            || assets |   
============++========+===
 2021-07-31 ||      1 | 1 
 2021-08-01 ||      1 | 1 
 2021-08-02 ||      1 | 1 
 2021-08-03 ||      1 | 1 
 2021-08-04 ||      0 | 0 

[Xitian9]

Is the ideal behaviour to have it go to 1 day before or to the end date?

[simonmichael]

1 day before, consistent with everything else, for now.

…

On Aug 2, 2021, at 03:57, Stephen Morgan ***@***.***> wrote:  Is the ideal behaviour to have it go to 1 day before or to the end date? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[simonmichael]

Fixed by #1637.