Re: All changes but huffman encoding integrated + a question

The problem here occurs when one is evicting what one is adding.

This can be handled without ever using more memory that allocated to the
compression context. One can also deal with this by copying just the name
(as opposed to the whole name+value-list), if one is willing to trade size
for complexity.

A ring buffer *is* a linear memory model-- it requires only a linear range
of memory. When I talk about a linear memory model without heap
fragmentation, this is what I am talking about.

-=R
On Oct 15, 2013 8:35 PM, "Martin Thomson" <martin.thomson@gmail.com> wrote:

> On 15 October 2013 19:25, Roberto Peon <grmocg@gmail.com> wrote:
> > 2) Make this illegal; This makes the encoder more complex as the encoder
> > must attempt to do this encoding and then check to see if actually fit
> every
> > time it does this kind of header representation.
>
> Not to want to contradict Tatsuhiro, but I thought that you argued
> pretty strongly for this one.  Something about a linear memory model.
> I got the impression that this was OK with people, mainly because it
> was perfectly consistent: evict, then add.
>
> Did something change?
>