[plt-scheme] Re: Help Desk

From: Robby Findler (robby at cs.uchicago.edu)
Date: Tue Nov 11 14:41:29 EST 2003

At Tue, 11 Nov 2003 13:42:58 -0500, Jim Witte wrote:
>   For list-related administrative tasks:
>   http://list.cs.brown.edu/mailman/listinfo/plt-scheme
>    What annoys me with the current internal help desk, aside from the 
> "Getting data from server" message (I'm not on-line, there is no server 
> other than the internal httpd process),

This is the result of a poor abstraction that we have inside Help Desk
(and the web server). There is no httpd "process" involved, actually.
Instead, the PLT web server is invoked, but with a fake set of tcp/ip
primitives that do their work via internal pipes rather than tcp/ip
connections. Both the html renderer and the web server are tricked with
these fake tcp/ip prims and that's why you see those messages.

Hopefully, we can improve this at some point.

> is that I remember back two or 
> three versions of PLT scheme when there was an internal browser that 
> was at least twice as fast as the current one (I'm running the MacOSX 
> build on an 800 mhz G3 now, I was running on a 300 or 400 mhz g3 then I 
> believe).
>    I actually *didn't* like the external, JS-enabled (I think) browser - 
> it seemed slower than both the old internal browser and the current 
> internal browser - a lot of that was because the speed of the JS was 
> dependent on the web browser implementation.

Well, I'm surprised by that. I'm not sure why it would seem that way;
most of the code is the code from that era. Still, something is fishy.
We'll try to figure it out. Can you say specifically what feels slower?
Is it searching? If so, is it only the first search, or subsequent
searches? The html parser is different, actually, come to think of it
-- are there particular pages that render slowly?


Posted on the users mailing list.