> On Jan 24, 2016, at 9:14 AM, Jesse Olmer <[email protected]> wrote:
> 
> On Sun, Jan 24, 2016 at 7:19 AM, John Ralls <[email protected]> wrote:
>> 
>>> On Jan 23, 2016, at 10:30 PM, Jesse Olmer <[email protected]> wrote:
>>> ImportA has an automatic best match to Trans1. I write into the table
>>> Trans1's guid as the key, and 1 as the value
>>> ImportB has an automatic best match to Trans1. I update the table
>>> Trans1's guid as the key, 2 as the value
>>> The User changes ImportA to match Trans2. I update the table twice:
>>> {Trans1, 1}, {Trans2, 1}
>>> 
>> 
>> Hmm. ISTM that's going a bit beyond what the bug asks for, but maybe I'm not 
>> thinking through the situation fully. How are you going to present this 
>> information to the user? Are you going to change the matcher logic to take 
>> it into account, so that if there are (say) 3 similar existing transactions 
>> and 3 imported transactions that match them with equal scores the matcher 
>> will assign one of the imported transactions to each of the existing 
>> transactions?
> 
> I wasn't planning on changing the match heuristic at all. This is
> simply the best way I could think of to solve according to "Possible
> Solution 2" in the bug description. Copied here for convenience:
> 
>> 2) Indicate (either by foreground or background color, or by including a new
>> column with a "cleared" indicator) in the SMET window that a transaction
>> has already been cleared by matching during a previous import process.
>> (Potentially this could be expanded to indicate that the transaction has
>> already been chosen for matching during the current import process as well.)
> 
> My change indicates previously cleared transactions by adding a
> "Reconciled" column to the match picker and populating it with
> gnc_get_reconcile_str. For the portion of the solution in parenthesis
> (indicate that it's chosen during the current import) the solution
> we've been discussing was my attempt at that.
> 
> The use case here, as I understand it (and have personally
> encountered) is that the import tool provides no help when trying to
> match 2 identical transactions. Both imports will auto match to the
> same transaction, and you'll be left with 1 of the 2 transactions as
> not cleared. Adding a column to match picker which indicates pending
> state helps the user to ensure that the two identical imported
> transactions are distributed evenly to the two identical transactions
> on the account, and everything ends up fully reconciled.

OK. For that purpose the GUID GHashTable will work fine.

I'll wait to see the code before I comment on the UI, but I'm worried about it 
being confusing.

Regards,
John Ralls


_______________________________________________
gnucash-devel mailing list
[email protected]
https://lists.gnucash.org/mailman/listinfo/gnucash-devel

Reply via email to