Each day I browse SO, there's a fair share of questions asking for a magnificent wonder of code conversion - from Java to JavaScript, from Perl to PHP, from SQL to Haskell etc.
Usually it means the OP got some legacy application and was asked (as politely as possible, of course) by their bosses to rewrite it into something more shiny/less crappy ASAPOES.
In extreme cases, some 100+ LOC monster is pasted directly into the question, making it as unhelpful for anyone but the OP as possible. That's rare, but that's really annoying, have to admit.
The problem is, I don't know how to deal with these questions correctly. The most natural reason for closing them was 'too localized', because they usually (but not always!) are. But that reason is gone from the list now.
There's, of course, 'minimal understanding of problem' check - but, frankly speaking, I don't think this reason will give a really good explanation of what's wrong with those questions. I'd rather see something more direct, but fail to come to any good AND concise line explaining what's really wrong with that 'conversion' approach. Are there any good ideas about it?
P.S. Also, there's another type of 'conversion' questions - when someone shows an idiom/pattern taken from one language, and asks about how better to express the same ideas in another language of choice. While these are on the edge on being too subjective, I suppose, this particular question is not about them - it's mostly about 'gimmethecodez' ones.
UPDATE: Ah, now I see this question among the links, which is basically about the same problem. Ironically, it has 'too localized' option choice as an accepted answer.
UPDATE 2: Reopening the question, as the 'minimum understanding' reason has gone away (here's the announce), and I'm really not sure about how the remaining ones fit in this case.