The days are shortening markedly. The sky straight above is a dark bluish indigo when I get up. And mist is hiding the horizon which should be a dark red. It's 11°C thanks to the open sky and thanks to the sun driving away the mist we have an open sky after 9:00 as well.

During my search for the trigger file problem I find a bug in the invoicing program. Last year and this year we have been plagued by duplicate record in the statistics file. After searching form more than a man month I finally found, and solved, the problem. Now I found a potential second source for the problem.
Before an invoice is made the invoice must be calculated. the invoice program checks in the trigger file for the calculation flag, if the flag is down, no invoice is made. If the flag is not down the invoice is made.
When the trigger isn't found the flag is not down and the invoice is made.
This is not correct of course because no trigger means nothing has been calculated (or, even worse the invoice has already been completed).
This could lead to duplicate statistics (and other nastiness) but until now nothing has actually gone wrong. That is been reported as such.

And I weeded some more in the PRS.

And rather late (about 17:40) Robert from Belgium calls about the correction for the archived Val problem I solved a few weeks back. He asks for instructions (just do what is described as solution in the problem report. Duh ) and then ask how he can check if the correction program worked as it should. That last is tricky. To be honest I don't know the answer. because I don't know how the problem records actually entered the system. I can only suggest to check the orders and related invoices before and after the correction.
Right he says. and mails me a user profile and password so that I can check.
Tomorrow.

Adios