Account matching on transaction import does poorly after account restructuring

[John Ralls]

> On Feb 13, 2016, at 4:42 PM, Jim DeLaHunt <from.gnucash at jdlh.com> wrote:
> 
> Hello, GnuCash users:
> 
> As mentioned in a couple of recent posts, I recently began a spurt of transaction importing, after a long pause. I got re-acquainted with the importer for OFX transactions, and its Bayesian matching capabilities.
> 
> It wasn't working well for me. Even after importing six months of OFX files from my credit card company, the matcher still wasn't able to figure out the correct account to use.
> 
> I did some digging. I made major changes to my GnuCash account names and hierarchy last year. I believe I'm paying a penalty now for that.
> 
> This is captured in bug 745896 "Bayes Import Map does not use GUID but account name, leading to outdated entries." <https://bugzilla.gnome.org/show_bug.cgi?id=745896>.
> 
> Has there been discussion of this issue beyond the bugzilla entry?  If so, I'd like to read up on it. I don't see anything from searching the list archives or wiki.
> 
> I have some ideas for how to implement the fix suggested in the bug report (use account's GUID instead of account's name as the identifier). I'll start a thread in gnucash-devel about this, and record any conclusions in bug 745896.
> 
> Inspired by a Perl script in the GnuCash wiki article "Bayes" <http://wiki.gnucash.org/wiki/Bayes>, I wrote some XSLT stylesheets which read an uncompressed GnuCash XML file and:
> 1. list the matched accounts in the Bayesian import maps
> 2. prune a specified set of accounts from the Bayesian import maps
> 3. completely remove the Bayesian import maps
> 
> I plan to post them in Wiki pages referenced from "Bayes", in the spirit of the Perl script there, unless someone has an idea for a better place.  I'll also put references to them in bug 745896.
> 
> There, now there is an entry in the -user list archives pointing to this bug report, for other users to find when they have the same problem.
> 

Robert Fewell said something about looking bug 745896 after https://github.com/Gnucash/gnucash/pull/45. I don't know if he's made any progress.

BTW, no doubt your XSLT stylesheets will be an improvement over the perl script, but that PR provides an editor for deleting matcher keys, both Bayes and non-Bayes, in the GUI for the next major release. They're unrelated to bug 745896, so please don't add noise to the bug report.

Regards,
John Ralls