[REBOL] Re: Rugby performace starting to fade
From: petr:krenzelok:trz:cz at: 19-Feb-2002 20:47
Maarten Koopmans wrote:
>Hi Petr,
>
>I don't want just positive reports, but the thing is that I was pretty sure
>that it was not Rugby, and well, I am sort of proud of it, so I kind of....
>well you read it.
>
>You are right, I was not nice at least, sorry....
>Did you win with ice hockey ? ;-)
>
No, we saved our win for the final match ;-)
>I don't get why you have to change things now.
>Rugby/http uses (and always has) rebols standard read, so...
>
hmm, I will try once again tomorrow. I don't know why is bypass ommited
if I have "127.0.0.1" specified in my user.r file. However - I got one
error with Rugby today, and the port output showed host number of my
local IP, but not 127.0.0.1. But I am not 100% sure now, so I will try
to test tomorrow once again.
>I don't know what is going on. System administration in your company?
>[I know we suffer from that sometimes]
>
Yes, I know. I was just suprised that I ran four interpreters from the
same directory, using the same config user.r file.
>httpr simply does open/copy/close, so...
>
I will try once again tomorrow morning ....
>I don't understand Holger's reply regarding Rugby's architecture. Don't they
>advertise with the fact that all scripts run the same on all platforms?
>
Well, it was not like that. From what I remember Holger was saying
something like there is no difference in TCP layer inside Rebol, so it
has to be a) Rugby b) OS. And I think it is definitely OS. W9x has imo
delayed ACK, and if Rebol is internally sending only one packet at a
time, OS waits a) till the timeout period runs out b) till the second
packet arrives .... and only then sends ACK ...
But I can be wrong, so don't draw any conclusion yet ... It would be
interesting though, if more folks could run proposed Rugby test using
W9x systems ....
Cheers,
-pekr-