Re: Compact Data

From: pinkfloydhomer_at_[hidden]
Date: 2001-03-18 17:39:15

--- In boost_at_y..., Gary Powell <Gary.Powell_at_s...> wrote:
> 
> > > Is there currently in standard C++ a structured way to 
represent 
> > data 
> > > compactly, other than the old C bitfields?
> > > 
> No.
> 
> > > If there isn't, couldn't this be a relevant thing for Boost?
> > > 
> Because physical data storage is a machine dependent problem, its 
not a very
> good candidate, unless you use say, compression which while not 
having
> identical physical formats, does have identical representations in 
code.
> 

It doesn't have to have anything to do with physical formats. I'm 
just talking about the ability to store at least the simple data 
types in some structure that is packed optimally, even if the access 
effeciency is worse. Why can't I say: I want an unsigned integer of a 
certain size, either by stating the number of bits, or even better, 
by giving a range: from 100-200; a floating point number from -1.0 to 
1.0, a signed integer from -16000 to 16000 and two bools packed in 
some datastructure?

I can easily implement this myself on a given compiler, but I 
shouldn't have to.

> > Not to be rude, but isn't anybody but me interested in this?
> I have been in the past, but I kept finding that the code space 
which
> expanded using bitfields out weighed the data space savings. (Just 
my
> problem domain though.) 

And in my problem domain, the space savings are critical. It's 
basically a dynamic programming scheme where the results of an 
exponential search of a node in a game tree, is saved in a table to 
gain access to it later, if the same node should occur by 
transpositions or otherwise. In my domain, 128 MB is easily utilized 
for this, if available. The more the better. So whenever I can 
squeeze my data to half the size, I just saved buying another 128 MB 
RAM module.

The point is that we can't possibly foresee all the uses of such a 
concept. I think it is obvious that there are problems that are 
solved best if space usage is effecient.

> From the lack of response it doesn't look like there
> is anyone else in boost with this as their current focus.)
> 
> Boost is a self directed group. If you have a strong interest, the 
rest of
> us are interested in what you are interested in. However we also 
have other
> interests that keep us busy. It's a great place to meet with other 
library
> writers, but there are no guarantees that you will get any direct 
help,
> other than question/answer support, and code reviews.
> 

Of course. I was just trying to get ideas from others who had been 
thinking the same thing, not being the most brilliant library 
designer myself, to say the least. I'm surprised no one can see the 
general use of this, or is at least interested. 
People are spoiled with memory nowadays :) Most people's problems 
don't care if there is 32 MB or 256 MB available. Some do, though.