reg -M each ancestor account with posting duplicates descendant amounts

[jolmg]

How to reproduce

hledger -f - reg -M --forecast=-2020/02/01 <<EOF
2020-01-01
  (a)        1.00 USD

~ every day
  (a)        1.00 USD
  (a:b)      1.00 USD
  (a:b:c)    1.00 USD
EOF

Expected

2020-01                 a                                 31.00 USD     31.00 USD
                        a:b                               30.00 USD     61.00 USD
                        a:b:c                             30.00 USD     91.00 USD

Actual

2020-01                 a                                 91.00 USD     91.00 USD
                        a:b                               60.00 USD    151.00 USD
                        a:b:c                             30.00 USD    181.00 USD

Further notes

Total is expected when omitting -M:

$ hledger -f - reg --forecast=-2020/02/01 <<EOF
2020-01-01
  (a)        1.00 USD

~ every day
  (a)        1.00 USD
  (a:b)      1.00 USD
  (a:b:c)    1.00 USD
EOF
...
2020-01-31                       (a)                       1.00 USD     89.00 USD
                                 (a:b)                     1.00 USD     90.00 USD
                                 (a:b:c)                   1.00 USD     91.00 USD

Issue also present when using -W:

$ hledger -f - reg -W --forecast=-2020/02/01 <<EOF 
2020-01-01
  (a)        1.00 USD

~ every day
  (a)        1.00 USD
  (a:b)      1.00 USD
  (a:b:c)    1.00 USD
EOF
2019-12-30W01           a                                 13.00 USD     13.00 USD
                        a:b                                8.00 USD     21.00 USD
                        a:b:c                              4.00 USD     25.00 USD
...

Platform

OS: Archlinux
hledger version: 1.21

[jolmg]

Workaround could be to rename accounts of postings to ancestors to something distinct:

$ hledger -f - reg -M --forecast=-2020/02/01 <<EOF
2020-01-01
  (a-w)      1.00 USD

~ every day
  (a-w)      1.00 USD
  (a:b-w)    1.00 USD
  (a:b:c)    1.00 USD
EOF
2020-01                 a-w                              31.00 USD     31.00 USD
                        a:b-w                            30.00 USD     61.00 USD
                        a:b:c                            30.00 USD     91.00 USD

This way the totals between using and omitting -M match.

[jolmg]

I noticed that bal -M doesn't add-up the descendants in the ancestor entries:

$ hledger -f - bal -M --forecast=-2020/02/01 <<EOF       
2020-01-01
  (a)        1.00 USD

~ every day
  (a)        1.00 USD
  (a:b)      1.00 USD
  (a:b:c)    1.00 USD
EOF
Balance changes in 2020-01:

       ||       Jan
=======++===========
 a     || 31.00 USD
 a:b   || 30.00 USD
 a:b:c || 30.00 USD
-------++-----------
       || 91.00 USD

unless you add --tree:

Balance changes in 2020-01:

       ||       Jan
=======++===========
 a     || 91.00 USD
   b   || 60.00 USD
     c || 30.00 USD
-------++-----------
       || 91.00 USD

Haven't looked at the source yet, but I wonder if --tree activates something for bal that is active by default with reg. Maybe this could be just a matter of toggling a boolean somewhere.

[simonmichael]

Thanks for the report! This seems to have broken somewhere in 1.19:

~ every day
  (a)        1.00 USD
  (a:b)      1.00 USD

$ hledger-1.18 -f b.j reg a --forecast=thisweek -W
2021-06-07W23           a                                                      7.00 USD      7.00 USD
                        a:b                                                    7.00 USD     14.00 USD
$ hledger-1.19 -f b.j reg a --forecast=thisweek -W
2021-06-07W23           a                                                     14.00 USD     14.00 USD
                        a:b                                                    7.00 USD     21.00 USD

[jolmg]

👍 Though, I think you mean after 1.19? 1.19's output seems correct. Or maybe you pasted 1.18's twice?

[simonmichael]

Indeed, I messed up - I have edited the comment now.

[simonmichael]

Related: https://groups.google.com/g/hledger/c/zyMoqMzg1Fc/m/8G5hSGSDCAAJ

[Xitian9]

Git bisect says this one was my fault: 30f02b0. I'll look into what exactly broke.

[simonmichael]

Thanks @Xitian9 ! Fix merged.