I understand what the point of these two polls were, and I would have been interested in the results, but I have some serious problems with the way these were presented, which I think makes the result kind of wrong:

Let's start with the first poll. The wording here was pretty bad I think (and that's probably due to the lack of native-english speaking individuals who produced it).

I think the question should perhaps have been:

"Should Haiku also release GCC4-Only builds?"

Which may seem similar, but it makes all the difference I think. The way it was phrased seems to ask: "Should Haiku only release GCC4 builds?". Was that really the intent of the question? Because that basically suggests throwing out BeOS binary compatibility completely if you answer "Yes".

The second question is even stranger... as it provides a 3rd answer which is totally unrelated to the question.

The answers should have been yes or no. "Sometimes" doesn't matter - and what does that even have to do with GCC4? A BeOS app could never be built with GCC4, and a Haiku app built with GCC2 isn't necessarily BeOS compatible. It seemed like a very loaded question and a bit biased.

In any case, I didn't vote on either because I felt the questions were poorly thought out and misleading to the poll-takers.

FWIW, I do usually build GCC4-only images for my Haiku testing anyway - as I hardly ever require BeOS binary-compatibility... If I had a choice, I would probably recompile all the software I used in GCC4 anyway (assuming source was available)... I don't mind if the API/ABI changes in the future.