When I get up at 5h45 it is softly raining but by the time I hit the road to work it has stopped and the roads are steaming, returning the rain whence it came. Leaving a kind of thin mist that slowly rises, greying out the sky. Whatever goes up must come down and in the afternoon the water is back. As rain.

The correction program works but testing is not at all easy because we cannot recreate the actual, error condition that must be solved. And because there is no margin of error allowed. This is a program that can only run once. The only way back is a database restore. The problem is that to verify the correctness of the program it must run before the backup and the result can only be checked after the backup. Thus restoring would cause the loss of a full day worth of orders (and unhappy clients) or the running would require a double backup which doesn't fit in a single night. Running over a weekend would be the best option but the opco's are not willing to call in staff for a weekend to solve a problem we created.

One dart and bulls-eye is required.

 

Did a third round of spraying the potatoes. The spraying did stop the blight so maybe it is the last round.

---

Still waiting for 38-hour days.

No rain and only a little bit of mist (at 14°C) in the morn'n turning sunny and quite warm soon enough. With all the sun and the humidity there are bound to be, localised, thunderstorms.

Still testing and changing the correction program. With a few assists to Ronny thrown in.
Yep, fairly standard day.

 

Mailbank seems to be having a serious problem at its hands. During the week I have my mail forwarded to my account at work and in the weekends I switch that forwarding off. So, last Friday I switched it off. The switch did proceed but the status page at their web site did not register the change. That has happened before so I don't really care.
Sunday evening (around midnight locally which should be in the early evening on the Mailbank servers) I switched forwarding back on. Yesterday morning I was not getting any mail so I checked. The status page remained unchanged, showing the redirection change of 26^th. The forwarding form does show forwarding to be active and their server status page claims a smooth recovery from a power outage on Friday as last event.
Yesterday evening I found all my mail at home so redirection was not working. I switch it off on the redirection form and back on again half an hour later. The status pages don't change.
Today the redirection still isn't working so I get all my mail at home, after 20h00. I haven't lost mail (I think) but I don't like the situation. Mailbank didn't react to my 'complains' (softly) yet. Maybe with Independence day passed they come back into the office. Probably to "press ANY key" on a console or something in that vein.

So if you sent mail just be patient, I get round to it in the evening. If you expect an answer soon you'd best use my office account (qjsw @ oce;nl).

 

Did I mention the chance on some thunderstorms?
On the way home, on the highway, I met up with one. Visibility dropped from normal to about 10m. Like driving into a wall of water.
Coupled with rather dense trafic at around 120Km/h is not funny. Especially with fools slamming on the brakes and swerving off to the right without being able to see if there is free space at the right. I didn't hit anything and I wasn't hit by anything (I did 'jump' to the fast lane and slowed down gradually to a still unreasonable 80Km:h). Wizards luck?
Typical for this type of shower it rains harder than that the water can run off the road so in no time you are driving in a few centimeter of water but about ten kilometer further you're back on dry road and in the sunshine.

---

Life may look boring, untill you realise it is a game.

Unlike yesterday it starts out dry. For the rest it an exact copy, dry and sunny but cool for summer (14 to 21°C).

I lost yesterday. I am sure I did put it in its place on Monday but some how I lost it and I just cannot find it anywhere.

 

One thing I remember, a new problem. Not that I needed such a thing but ... well.
All orders remain in status VAL in the Integration test environment, which sits between the development/test environment (programmers playground) and the production environment (from which distributions are made). In theory the managers must test in this (and the system test) environment. Of course Jan, when he tests does that in the development env.
I spent time yesterday on it and the whole morning today. Stepping trough the source with a debugger (which is all but simple for a process that gets submitted from another system) did not turn up any clue. After sifting trough several hexadecimal dumps I finally succeeded to reproduce the problem in the development environment. And solve it in 15 seconds.
In the new release, where the error occurs occurred :) the order header file is two bytes longer than in the old release. The program running in the Integration environment was compiled with this new length but an internal buffer was still defined with the old length.
I started on a redesign of the program some time ago but this got pushed down on my priority to do list. The redesigned program works correctly but is not yet usable (it has an endless loop somewhere else) so I made a temporal fix (copy source, change buffer from 251 to 253, compile and walk the distribution carrousel).

After finishing that Iwan, from Norway, reported a problem with activating of contracts. Well actually the activating of a specific contract.
Standard procedure.

• First I try to recreate the problem here. Doesn't go, I mean it goes. Well the problem doesn't show up and the program does work as expected.
• Next I ask my manager. Oeps, he is on holiday for three weeks.
• Third, because the program works over here, I check the remote system. Do they have local adaptations in our programs? Did they cancel jobs? Do they have the defaults sensibly? that sort of things. Oeps, I haven't got a user profile/password on their system.
• ...

This is going to take some time.

---

Of course life is a game without an undo and where failure is not an option.

The day evolves from drizzle (14°C) to various shades of gray to light rain and back to grey.
Yep, that is a Belgian summer in the making.

I received a user profile and password for the Nordic system so I log on and check the problem. They reported that they couldn't get a contract activated. I check that contract and it is active, without apparent problems. A few messages later I get the correct problem report. They wanted to change the prices on the contract and couldn't get the new prices active.
Aha
I check all the files involved and find the situation as it should be before the running of the activation program. So all looks normal. I inform the user that he should run the activation program but later (in the job logs) I find out that he left early. So I run the activation program myself. Well it run without errors but when I check later I notice that the contract is not updated !! So there is something really wrong with the program. The Contract change requests are updated though so I know where the program makes an error but it is in a place where it shouldn't (it doesn't find a record but I can see that very record, displayed on screen). The only way to get the exact problem is running the program trough the debugger. The user profile I got is for a normal user so I don't have access to a command line (which I can get around to) but I also don't have authority to the debugger. Can't get around that.
And, of course running the program the first time has used and altered the request records so I don't have live data anymore. And I cannot just make a guess at it (well I can but I can't ..) because this is the production environment. Any test I do ends up on an invoice to a real client.

All this checking is done using a telnet session on the AS/400 in Norway which shouldn't be a problem. Only I have to wait several minute after each command I enter (for example I started a query resulting in 6 records and got the result back some time after reading an article on Linux World and I am not at all a fast reader). That is not normal but I guess it has something to do with the communications module on the Nordic system because after 17h00 things go much smoother (wait times of about a minute).

Rain in the ... ah no sun. Heu, grey, sorry blue, oh, rain. I mean sun. Well now ...
Sh*** let us just call it unsettled. Basically lots of rain and greyness, a sad day. The evening is nice but cool.

Shopping. Food, wine and a gigabyte worth of Zip disks (4x250).

Late in the morning I go shopping with my brother for some wood; he is building a greenhouse. He needs me because I have got a Fiat Ulysse (19DTS from 1995) (ps the fiat site has been changed but not improved recently) which is a monovolume car with more than enough space while he has got a Toyota Celica which is a coupe. And how do you transport a laminated wooden plate of 2.4 by 1.2 meter? It fits easily in my car but with his sports car he can forget it;

Sad day? yep, I spent some time crying. after shopping, just because I am very sensitive.
No duh, not sentimental, I am just making onion soup. My father may not eat onions (too much potassium which could cause problems with his kidney-dialiysis) so it is quite some time ago I had onion soup, which happens to be my favourite. Father and Suzan go on vacation till Monday so I can prepare and eat what I want.

Most of the day is rainy so I take it easy. I picked up some old (1911, 1912) almanacs (calendar plus jokes plus advice plus saints plus ....) Apart from the spelling and number the jokes haven't changed much. They just turned a little bit more explicit.

The evening is fine, dry with blue skies but cool. Not ideal for barbecuing but not too bad either, enjoyable. Cracking a good bottle of wine makes up for the weather.