On Tue, Aug 22, 2000 at 05:42:18PM +0100, Theodore Hong wrote:
> It feels like a bad design choice to arbitrarily set a fixed upper limit to
> the size of data. ("No one will ever need more than 640K of memory...")
> Why do we need one?
Ultimately maybe, but for now we are just making calls for this version,
and we know we will have to go through the protocol once more.
Remember that we are going to be using file splitting. Most people agree
that we want 128 K or smaller parts - so puting a 200 meg limit would be a
hundred fifty times normal. Yeah I know, any fixed number is too small, no
matter how big it is... but still...
> If you don't have enough temporary space to store a reply, can't we fall
> back to streaming it directly from the in-socket to the out-socket (the way
> we did in the beginning) instead of writing it to disk?
That code was a little bit of a cludge, and most for the current code for
validating the data was written with the assumption of middle step of
writing the data to disk. And what do we do if we don't write to disk and
get a RequestFailed?
It can be handled, best via a circular file like I described, but it's a
bit of a bitch, so I would rather have a limit for now.
>
> theo
>
>
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev at lists.sourceforge.net
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev
>
--
\oskar
_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
