I've just seen this question, PHP Hash Removal, where the problem was that
str_replace("#", "", $foo) failed; and the correct answer is that it should use single quotes instead of double quotes. OP says it worked.
I'm glad it somehow made the trick for him. However, I'm almost 100% positive that PHP's
str_replace function doesn't make any fundamental difference between single quotes and double quotes.
I'm afraid this kind of answer to a question, even though accepted, enforces programming by superstition, especially since no one could reproduce the problem on their computer (both kinds of quotes have the same, correct results for everyone). The guy who answered said he had the problem once and doesn't really recall the circumstances.
My belief is that the guy changed something else and didn't test it, and then when he was prompted to change the single quotes to double quotes, it suddenly worked because of that other change, or something along these lines.
There is also a certain number of other questions with very lowly scores being accepted, which seems as wrong to me.
What should be done about questions with accepted answers that visibly are superstitions? Should Stack Overflow be proud of leaving this kind of answer for posterity?
inb4 "With PHP you never know"