Daynote ,Svenson

MM-lxii     Thursday

2000-03-02

We have cooler weather again, just -0°C. The problem is that it is raining lightly. Problem? Yep rain on frozen ground makes driving a real adventure. Much more exhilarating than Lara Croft or Doom, or Unreal. I pulled out of two tailspins (one fast), while a police car in front of me did not succeed to pull out of a similar situation. Hehe.
The rest of the day was unremarkably gray with lots of wind.

The PCA stuff is compiled, tested and ready. Now I must find some spare time to test the unspecified small window version of it. Testing will be simple, finding the time is something entirely different.

Wilbert was trying to create a batch file to stop a bunch of database servers on our NT box. We typically have 5 or 7 or 10 or ... server jobs for the various working environments running. Often, when we have a problem on the development client (Ronny's PC) we must stop and restart the database servers. This takes no more then about ten minutes all included but while we are working with the V9 of progress it sometimes happens several times a day (four times in quick succession this morning, wow). So Wilbert wanted to put the whole action in a single batch file, not to speed up the process but rather to make it easier to do, without requiring attention. He is not proficient in DOS however so he typically asks me for advice. I noticed however that my knowledge of batch programming is a bit rusty.

The problem was that, to stop a server you must call a tool from the Progress Database. This executes as expected but somehow it doesn't return control to the batch routine. The same command has to be executed for each server. Tom finally suggested to use the RUN statement to start the tool (an .exe file) and that solved the problem. I was experimenting with COMMAND to start a new command processor for each action (good old DOS-3 technique) but I didn't remember the correct switches to stop the secondary processor after its command(s) were finished.
Another thing that I am not certain about is if the batch language behaves the same under NT as it did/does under DOS. (I did notice that long filenames were not supported but further than that ...)

And I spent a lot of time 'correcting' some adaptation from Koen. The problem here is that Koen is a very good programmer but he doesn't play by the standards.

When we make a change there is always a problem report relating to it. The rule is that you log, at the top of the source, the date, id and PRS number (and a one line description) of the change and that you mark every changed line with the id. Another rule is that you never actually remove the original code, you convert it to comment line(s) and add the changed source under it. The intention of this is that opcos with a local variation can easily find the changes and apply them to their localized version.

Koen doesn't comply with these rules, he just replaces the code where it needs to be changed. If everything works, and he is good enough that it mostly does work first time around without testing (which he doesn't do enough), this is not a big problem. This time he had made a few mistakes. These were inhibiting some testers but he didn't have time to quickly solve them ( ). So I had to solve the problem, with the help of Frank. While I was in the sources I added the necessary comments and I added the old code (commented out) as well.

The result is that I spent less time on my own projects than I should. And I am down on fuel.

Did any one notice I had moved Wednesday back to January :-)

Adios