Pnina, regarding reproducible bugs, I can perfectly understand e-on. I was a programmer, you were a programmer, so we have similar life experiences in this respect, it's interesting how we see this situation in a completely different way. :-)

There are several general tools/processes available for a programmer to find a bug. By far, the easiest is to reproduce the situation and execute the code step by step in that area until (s)he finds the culprit line. Generally speaking, if you can't find something, how can you fix it?

If the bug is not reproducible at all, this could mean many things. It could mean an influence of the user's particular system in the program, it could mean that the user didn't exactly transmit the programmer all the bug context (like steps he did before or settings he changed), it could mean a non-initialized variable (producing random effects, or no effect depending on what was done before), etc, etc. In these situations, it's theoretically possible that the programmer double checks the code line by line but this, depending on the situation, can be close to impossible (it's impossible to double check thousands or millions of lines).