Other QIF bugs (or misfeatures)

Derek Atkins warlord@MIT.EDU
14 Feb 2001 14:57:56 -0500

grib@gnumatic.com (Bill Gribble) writes:

> On Wed, Feb 14, 2001 at 11:32:43AM -0500, Derek Atkins wrote:
> > Well, I can give you the QIF file that this fails on, if you like.
> > Basically, I chose "Liabilities:Credit Cards:Chase MC" as my account.
> > After import, it generated "Liabilities 2:Credit Cards 2:Chase MC".
> Could you try the patch I just sent to -patches?  I think it should
> fix this situation.  Basically it's just smarter about account type
> compatibility.  You should probably remove the file
> ~/.gnucash/qif-accounts-map before you run the import, since it has
> some misleading info in it and needs to be blown away.

Sure, I can try it.  It definitely helps some things, but
unfortunately not this particular problem.  I blew away my
qif-accounts-map file, but after I load the Chase QIF I still wound up
with that same duplicate account tree, "Liabilities 2" ->
"Liabilities:Credit Cards 2" -> "Chase MC".

There are two clear bugs here.  The first is that it shouldn't be
building the secondary account tree.  The second bug is that the
"intermediate" account is actually named "Liabilities:Credit Cards 2"
instead of "Credit Cards".  Yes, there is actually a colon character
in the account name.

> > This does not happen to me.  If I select an account from the account
> > picker, the checkbox does NOT go away.  Indeed, the checkbox will even
> > appear at times when it wasn't there earlier!  I'm hoping this isn't
> > another guile 1.3 badness.
> Nope, just a Bill Gribble badness :) This was being detected all
> wrong.  Should be fixed now.

Yep, this is fixed.

> > I think the importer should only care about the leaf account type
> > and work upwards from the end, rather than downwards from the top.
> Well, I *was* being too strict, but Gnucash has real restrictions on
> what kinds of accounts can be children of other kinds.  It's important
> to make sure the user doesn't end up with an "impossible" account tree
> just because some names matched up, or because you told it to do
> something impossible which it didn't have enough info to know was
> impossible at the time.

One would think that if the account already exists, then you've
already defined what goes in the account, and the hierachy has already
been validated.  Therefore, you only need to make sure that the
transactions being loaded match the account-type already created.
You should only worry about the hierachy in the case of a new account.

> > Oh, one more thing that may be an issue: ticker symbols for
> > commodities.  If I'm importing a QIF file into an "already existing"
> > account, it still asks for commodity ticker symbols.  It would be nice
> > if it would just use what's in the account already instead of asking
> > my every month when I import my account statement.
> haven't fixed this yet.   You're right tho.


> b.g.


       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord@MIT.EDU                        PGP key available