Comments to IRC discussion: transaction importer and imbalance-account

Christian Stimming stimming at
Mon Jun 4 04:45:00 EDT 2007

A comment to

23:35:59 <lasindi> I am testing right now with CSV files that have  
just three columns per row: date, description, amount. First the user  
selects an account to import to, then for each row I create a  
transaction and add one split. I set the split's account to the  
selected account and its amount/value to the amount in the file. Then  
I hand these transactions to the importer.
23:42:19 <warlord> e.g., see import-main-matcher.h,  
23:42:38 <warlord> Looks like it DOES take a Transaction* BoO!
23:43:47 <lasindi> Hehe, yep

Right, that's ; see  
my explanation on how this could be fixed, but I consider this a  
slightly more involved task.

23:56:58 <lasindi> The problem is that these transactions go through  
the importer into the source account with apparently single splits, so  
there are no destination accounts for the transactions, not even  
23:57:38 <warlord> This is an importer bug..
23:58:44 <warlord> As I recall, the importer just doesn't add a second  
(balancing) split in the 'default' case. I consider that a bug. If you  
want to fix that you have my complete support ;)

Same from me - this behaviour of the Importer should be changed in the  
long run so that the Importer adds a split to the Imbalance-Account  
for all transactions where no "other account" has been selected. From  
the UI point of view, this probably doesn't change what the user sees:  
The green transactions have a other-account other than the Imbalance,  
and the yellow transactions have the other-account set to the  

However, internally this requires rather a lot of changes in the  
Importer implementation. For example, the current "flag" to  
distinguish the yellow from the green transactions is simply the query  
whether there is only one split vs. two splits. Also, currently if any  
transaction is committed (xaccTransCommitEdit), the other-account is  
saved for future automatic matching. If the yellow transactions are  
committed with the Imbalance-account as other-account, that account  
has to be excluded from the automatic matching, of course.

lasindi: In summary, I agree this is a bug and should be changed. But  
for now I would recommend you should ignore that importer bug and just  
focus on the CSV import part. Either you or someone else will probably  
pick up this issue with the generic importer and Imbalance-account  
handling sometime later...



More information about the gnucash-devel mailing list