Oh fun (geek thread warning!)

Well, I warned you.

Don't you just love it when you spend 2 days fixing a problem, pulling your hair out because everything you try, and you KNOW works in code, doesn't work when you press buttons in the application?!?!

Basically, using set theory to group items (easier if I don't go into too much detail probably) based on certain attributes those items may, or may not have.

The function should of returned a certain grouping, but wasn't. I tried tweaking the code, I even tried throwing the code out the window & using a completely different approach, nothing worked, so I went back to the original code to step through it & see what was happening.

I still couldn't make head nor tail of it last night, when I left the office 1 hour 30 late. I was also thinking about why the hell this wasn't working late last night. Basically, the entire company was waiting for me to fix this bug before it's released to a customer, and by this morning, we were starting to push 24hours behind the specified deployment date, which means charges are applied by the customer.

So I came in 1hour 30 early this morning to take a "fresh" look at it. Pumped myself up on coffee, and set about it.

Turns out, the application was configured wrongly, not in one database, but in TWO. Both test environments I had to play with were incorrectly setup. Nice. So the attributes that it was trying to compare based upon, didn't exist. So I changed the configuration, and it was still no better.

After looking a little deeper, I find that the actual attribute values weren't as they should of been for the test case. So I spend 45 mins fixing the data in various places, and re-run the test. Guess what happens! It works EXACTLY as documented. Bleeding thing!

To add to my mood this morning (frustrated is probably the tip of the ice-burg), I find I've had one of my bug fixes in the last few days failed. Why? Because the data didn't exist when the tester came to test it. What the? Do they need nappies? The data doesn't exist, that's not the fault of my code! Can anyone guess what I did with that Failed Bug Work Order?

Someones head will roll for yesterday, and if I keep getting bugs bounced back because of a lack of data, I'm not going to be a happy bunny! I've already talked to my line manager (called him on his mobile to wake him up ) about both bugs, and he won't be too happy that he's been woken up. Not really my problem though

I love it when a plan comes together. I get myself into similar situations interpreting data from spectroscopy measurments when I've made a compound in the lab. I spent all night a couple of weeks back deciphering stuff, and couldnt make head nor tail of what was going on; turned out the compound I was analysing displayed two forms in great amount of both at room temp (something called keto/enol tautomerism) which was difficult to spot as the peaks overlapped on the IR spectra.

More geeky stuff!

Matt

Sorry... got as far as 'The function should of returned a certain grouping' and got bored... hope it turned out nice in the end.

Originally Posted by Grinch

Sorry... got as far as 'The function should of returned a certain grouping' and got bored... hope it turned out nice in the end.

futher than i got

yup, know the feeling! Here we are currently arguing over how many test environments we need. I say keep it to a minimum but they want quite a few. *sigh*

Originally Posted by Kate

yup, know the feeling! Here we are currently arguing over how many test environments we need. I say keep it to a minimum but they want quite a few. *sigh*

You need enough test environments to be able to test everything properly, but quickly. You don't want to waste time setting up test environments.


At the moment, on my machine, there's probably 25 different test environments setup, each for a different version of the application, or a different customer etc etc. Then I have full access to support environments, including everything up to customers live databases.

Makes your head hurt at the best of times!

Sounds like alot of variables to account for, sure sounds harder than chemistry



Matt

[Luckypants]

Wish our developers were that thorough! Thier attitude seems to be 'support can fix the bugs' and sling the code out.

Well, I'm glad I came to work this morning! Remember those test environments I was on about? I had carefully set up (taking most of yesterday) a number of test documents ready for an Oracle update, I had discussed this fully with the developer yesterday as after I've finished with the documents he can then use them for his tests. I came in this morning to find the documents have been wiped as the developer decided that the database needed to be wiped ready for the Oracle update. (Front end database with an Oracle backend). Fantastic. Time to start again.

Originally Posted by Kate

Well, I'm glad I came to work this morning! Remember those test environments I was on about? I had carefully set up (taking most of yesterday) a number of test documents ready for an Oracle update, I had discussed this fully with the developer yesterday as after I've finished with the documents he can then use them for his tests. I came in this morning to find the documents have been wiped as the developer decided that the database needed to be wiped ready for the Oracle update. (Front end database with an Oracle backend). Fantastic. Time to start again.

I don't understand this at all. Why the hell would you need to get rid of test documents in order to wipe a database? If you want to wipe the DB, just "drop user <user> cascade;" If you want to free the space on the disk, then find the CTL files etc, and wipe those, then re-create namespaces etc.

Some people are just ijits!

(But you knew both the main points of this post already, I know this )