Date: 2001-02-27 13:49:28
Actually, neither of these is what I'm looking for.
Basically, (as a hobby) I am writing yet another shoot-em-up game. The
tried and true technique for managing bullets, ship thrust, bullet
holes etc. is to have a "ring buffer" or
"circular buffer" with a fixed maximum size.
You want a maximum of, say, 1000 bullets on the screen. As soon as the
ship fires the 1001st bullet, the 1st bullet "disappears". In terms
of the buffer, asssuming it has a fixed size of "1000". When you add
the 1001st entry, it overwrites the 1st entry.
The advantages are twofold:
1. You want to limit the number of artifacts anyway - having more
than 1000 bullets, for example, would slow down the frame rate
2. It is efficient because it never needs to allocate memory.
It's pretty easy to do - but I was hoping for an "STL-friendly"
version so that I can write
std::for_each(bullets.begin(), bullets.end(), draw_bullet);
std::for_each(bullets.begin(), bullets.end(), animate_bullet);
the idea being that g_bullets.capacity() would never increase unless
you explicitly called reserve().
So in the above example, I would want the push_back call to just
silently overwrite the oldest bullet, if the buffer was full.
Of course, a ring buffer like this is also useful for many network
protocols that keep "sliding windows" of messages that might need to
be resent etc. Most protocols define a "maximum" size of such a
window - again, any error would occur only if you couldn't find an
entry in the buffer at a later point, not when you are inserting a
new entry (and thus possibly overwriting the oldest entry).
These are the type of problem domains a ring_buffer or
circular_buffer would be used in.
--- In boost_at_y..., Howard Hinnant <hinnant_at_t...> wrote:
> There were two ideas floating around:
> 1. A circular_buffer that would extend its capacity when necessary
> (similar to vector and deque).
> 2. A circular_buffer that would throw an exception if the capacity
> Which one (or both?) are you interested in?
> Personally I'm only interested in #1, and am willing to donate the
> interface to an existing implementatation (taken pretty much from
> and deque). Unfortunately I can not donate the implementation
> but I would be happy to participate in and discuss design decisions
> such an implementation.
> boost_at_P... wrote on 2/24/01 6:30 PM
> >Hi everyone,
> >A while back there was a thread talking about a circular_buffer
> >class. Does anyone know what happened?
> >It would find it to be very useful.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk