Possible bug in assertions

[Craig Earls]

I can't decided whether this is a bug, or a misunderstanding on my part.

I have had a few cases recently where a balance assertion was acting strangely.  It appears that The postion of the assertion in the file stream is more important than its chronological position.

I am trying to generate a suitable toy example, but here is a word description:

If an assertion for a balance on 2012/12/31 is in a file that is included prior to a relevant transaction WITH AN EARLIER DATE. the earlier transaction is calculated as if it were after the assertion.

--
Craig, Corona De Tucson, AZ
enderw88.wordpress.com

[Gabriel Kerneis]

On Thu, Jan 03, 2013 at 10:44:41AM -0700, Craig Earls wrote:
> The postion of the assertion in the file stream is more important than its
> chronological position.

I think this has been acknowledged by John on this mailing-list in the past (but
cannot find it right now).

Best,
--
Gabriel

[Christophe Rhodes]

Craig Earls <ende...@gmail.com> writes:

> If an assertion for a balance on 2012/12/31 is in a file that is included
> prior to a relevant transaction WITH AN EARLIER DATE. the earlier
> transaction is calculated as if it were after the assertion.

So, this makes sense to me (believe it or not), and I use it as a
feature. My bank sometimes sends me credit card statements with a given
statement date, and then in subsequent months sends me new statements
with transactions with dates preceding the previous statement's dates,
probably because it has taken some time for a transaction to work its
way through the system, but once it has its effective date was the date
that I paid for something.

The transactions then go in ledger files in statement order, and balance
assertions work as I (but not you :-) expect them to work; on the other
hand, sorting by date gives the what-retroactively-actually-happened
story. (Given that transactions can have multiple dates, and that
there's no enforced policy on what the dates mean, I don't think that
there's anything else that a balance assertion can mean other than
through file position).

That's my thinking, anyway.

Christophe

[Craig Earls]

Isn't that what the effective date notation is for?

[thierry]

On Thursday, January 3, 2013 6:59:55 PM UTC+1, Gabriel Kerneis wrote:

On Thu, Jan 03, 2013 at 10:44:41AM -0700, Craig Earls wrote:

> The position of the assertion in the file stream is more important than its

> chronological position.

I think this has been acknowledged by John on this mailing-list in the past (but
cannot find it right now).

In this thread : https://groups.google.com/d/topic/ledger-cli/S1GdYc5J9Mc/discussion

Thierry

[Craig Earls]

Look at that, I even helped answer the question...Thanks thierry!

[John Wiegley]

Yep, assertions are done by ordering, not by time.

John