Daynote ,Svenson

MM-cxxxix     Thursday

2000-05-18

Night temperature is down as well (10C) so we are getting more typical spring weather lots of rain but sunshine too and temperatures closing in on the 20C.

I worked out a solution for the problem I found Yesterday.
The problem is that we have validations that can be switched on or off, but the client can also decide to run the validation online or in batch. If run online any error is reported via a message while the user is still talking to the client on the phone so he can ask/propose something to the client. When run in batch the order is not released to logistics but remains in hold, with a reason code. The client is not on the phone anymore so either the user solve the problem himself or he calls the client. One of these is intended to prevent entering the same order twice within a (user) specified period like 5 days.
To do this the new order is compared, on header level and line by line, against all the existing orders for that client in the period. If an exact match (on a subset of criteria) is found the order is assumed to be in error. Well it is just a warning because the order can still be confirmed an passed.

The problem is that, while entering the order (online) the real order files are not filled, all online test are run on a temporary file that only contains orders that are being entered. Once the order is saved it is added to the real files and deleted from the temporary files. Then the tests are run again. The testing program sees only one set of files, either the real files or the temporal files. Of course in the temporary files you never find a match.
So the same test running online or off-line gives different results.
There are several options but all of them result in a far reaching rewrite of the testing program. Which is not trivial.

No connectivity to be found yesterday evening or rather connectivity that would make a 300 baud modem look fast. So nothing got posted. Since the love bug cropped up I have had connectivity problems. I don't know if there is a relation, it could just be a tired server somewhere along the line. I must also admit that I haven't taken the time for retrying. If I didn't get a connection after two tries I just went away (to bed mostly).

At the moment Tom claims to have the highest performing box running *nix. But, considering this, for how long?
<BG>