CHANTING and DEBUGGING

Filed Under (work.BLOG) by WildFire on 27-10-2004

Time travels so fast.

It was exactly at this time when we arrived here yesterday from a tough client visit where I have to re-program two database applications on site.

Something that I rarely do.

Usually, every little detail is checked and compiled before I do the visits. If I re-program things on site it means somewhere along the line, I screwed up.

Imagine this... you've worked your arse out on one module for two weeks... sometimes even sleeping at around three in the morning. Polished, tweaked, tested and done everything on that module. Then at the site, when you're about to present things you realize that you haven't created a link to that module from the main menu.

I was calculating the number of ways and the angle of projection in case I opted to throw the monitor from the third floor of that building.

Good thing I was able to bring along the necessary files to fix and recompile things. Apparently it was linked but I forgot to set the form's ShowWindow properties to 'Show in Top Level Form' which prevented it from displaying that module properly.

That was problem number one.

The second problem encountered was from a different application which shows a little error that doesn't really affect the program but is still annoying which was apparently caused by a blank value in the program database configuration file.

Quite nifty since the error was able to pass two of my error handling mechanisms. I have to fire up the VisualFoxpro tool once again on site just to check the real cause of the problem.

The third problem was the trickiest.

(Sometimes I wish I could just choose treat and hand out chocolates every time some cybernetic super forces are doing this to me.)

The GITS database application (which was the first program I made for that client (which has been working smoothly for three academic years already)) started 'acting up'.

That application contains five table reports on one of its parts. Tables I, II, III and V works fine but the report on Table IV fails.

What makes the problem unusual is the Print Preview works fine. It shows all the records/statistics that is needed but when it starts to print things already, it shows only five records.

Sometimes six.

It even shows the report legend which partially rules out printer hardware-related problems.

I spent around an hour or so mapping out the steps I've been doing and the results. (And let's not even enumerate the permutation done on that part)

After a number of tries and failures and chants, it worked. It turns out that some blank statistics on the table screws up the printing part. Which is still quite mysterious since everything is good in the Print Preview part.

A copy of an older version of the program which I usually store in the server in case some problems come up, did help solve some problems.

But if you'd asked me, I still think the chant did solve the problem.

Ah... one of those days.