Re: Error when upgrading 1.2 db to 1.3 db

[Nigel Titley <..hidden..>]

On 09/01/12 20:11, Erik Huelsmann wrote:
> Hi Nigel,
> 
>>>>> Many thanks... I'll work through and find out which account is causing
>>>>> the problem.
>>>>>
>>>> Hmm, this fails at the first fence, with offset 0. I think its a more
>>>> fundamental error with the "account_save" procedure.
>>>>
>>> I'm still no further forward with this. Can someone put me in touch with
>>> the author of the migration script? I can't really move forward with
>>> testing 1.3 until we get an example database migrated across. And if we
>>> are having problems with the script, it's likely that someone else will
>>> have too.
>>
>> Could you send me a dump of the 'chart' table in your database? I'd
>> like to see what's happening here and what expectations of
>> 'account_save' in your table aren't met. We might want to add
>> pre-migration checks based on it, or we might want to change
>> account_save to compensate.
> 
> Looking at the code more closely, I think the migration code expects
> every account to be associated with a header of which the number
> alphanumerically precedes the account number.
> 
> Maybe that expectation is violated? If you replace the account_save
> function in the distribution with the one you'll find below, an
> exception will be raised if that expectation is violated.

I think this is probably what Chris is referring to in his workaround
suggested today. Certainly I have no headers in my accounts table apart
from the one the I added (at the end) for testing.

I'm about to add a header at the start, dump a new test copy of the
database and try again.

What is the reason for the header association constraint in the
migration code? I have to admit that I never saw any reason for header
accounts, but that is probably my lack of understanding

Nigel