1.19: Budget reports with --empty don't correctly handle unbudgeted accounts with boring budgeted parents.

[Xitian9]

Pointed out by @holmescharles in #939 [starting here]. Copied over here since it deserves its own issue.

~ monthly from 2019/01
    expenses:personal             $1,000.00
    liabilities

2019/01/01 Google home hub and home mini x2
    expenses:personal:electronics          $10.00
    liabilities                           $-10.00

$ hledger -f foo balance --budget -E -M --tree
Budget performance in 2019-01:

                               ||                       Jan
===============================++===========================
 expenses:personal             ||       0 [0% of $1,000.00]
 expenses:personal:electronics ||  $10.00
 liabilities                   || $-10.00 [1% of $-1000.00]
-------------------------------++---------------------------
                               ||       0 [              0]

[simonmichael]

I think this changed in 1.19 ? Should we class it as a regression ? (Ignore the layout bug in 1.19)

$ hledger-1.18 bal --budget -E
Budget performance in 2019-01-01:

                               ||                   2019-01-01 
===============================++==============================
 expenses                      ||  $10.00 [   1% of $1,000.00] 
 expenses:personal             ||  $10.00 [   1% of $1,000.00] 
 expenses:personal:electronics ||  $10.00                      
 liabilities                   || $-10.00 [   1% of $-1000.00] 
-------------------------------++------------------------------
                               ||       0 [                 0] 
$ hledger-1.19 bal --budget -E
Budget performance in 2019-01-01:

                               ||                           2019-01-01 
===============================++======================================
 expenses:personal             ||               0 [   0% of $1,000.00] 
 expenses:personal:electronics ||          $10.00                      
 liabilities                   || $-10.00 [   1% of $-1000.00] 
-------------------------------++--------------------------------------
                               ||               0 [                 0] 

[Xitian9]

It could go either way. The issue is with boring parent elision, and I believe prior to 1.19 there was no boring parent elision in multibalance reports (or budget reports). Basically, it used to be that --no-elide was assumed for budget reports, and so this issue didn't arise. It still doesn't arise: things are fine if you call with --no-elide. So from one point of view, it's a flaw with a new feature, not a regression.

However, the default behaviour did change from something which works in this case to something which doesn't work in this case, so maybe it's a regression. shrug

[simonmichael]

Yes it seems to be a 1.19 regression. @holmescharles, thanks and claim your https://hledger.org/regressionbounty at your convenience!

[holmescharles]

Thanks! I'm glad to have contributed to the project 😄.