On Sun, Jun 17, 2001 at 03:28:07AM -0400, Tavin Cole wrote:
> On Sun, Jun 17, 2001 at 02:53:07PM +1200, David McNab wrote:
> > Seriously though - I'll have a look, but I'm a bit sceptical about the
> > feasibility ot getting the required speed in a Java FProxy implementation.
> > Maybe some hard-assed coding decisions might result in being able to stream
> > MP3s (on a fine day, with the tongue angled 15 degress upward out the left
> > of the mouth, and no other processes running), but then again, maybe not. As
> > for MPEG streaming, I don't hold much hope.
>
> Where are you getting the idea that whatever observed slowness is due to
> fproxy being written in Java? My suspects are 1) slow network and
> 2) the fact that currently fproxy, even though it runs in the node,
> makes an FNP connection to the node to do requests.
FProxy *is* slow, but only when it hasn't been used in a few minutes and
the JVM or OS has filed its memory somewhere out of sight. After the
first couple loads it's not significantly slower than a C server.
--
"Is our system--was it invented by scientists?"
"No. If scientists had invented it, they'd have tried it out first on hamsters."
Mark Roberts | [EMAIL PROTECTED]
PGP signature
