>  PS> Server #1)
 >  PS> I could make a server SPECIFICLY for the BBS' different parts. Ie,

 > This is the right way to do it for OS/2.

Yeah, that's what I figured.

 >  PS> And also: How do I handle BIG requests? 

 > Either through the pipe or send a header in the pipe saying where the data
 > is on disk. Select which by a flag in the request. IE flag says 'if more 
 > than
 > 2k (64k), send as file if possible.' return packet either says 'data 
 > follows'
 > or 'data can be found at: c:\temp\data.$$$'

Yeah. I'm gonna have to find out how fast pipes is, I haven't used'em too 
much. I suppose copying 500 Kbs of data into a pipe is almost as fast as 
copying the file to disk. And just to make it easy I could make sure all the 
data is waiting in the pipe before reading, that way I wouldn't have to worry 
about the server being slower than the client (ie, trying to read a 10 byte 
data structure, and there's only 7 bytes in the pipe...). 

 > Start a thread for these, allowing the user to geep doing other things while
 > scanning. ie scan for new files while reading mail. When you click on the
 > 'new files' window, a popup says 'still scanning' if the thread hasn't
 > completed.

A VERY good idea, I hadn't thought of that. When handeling big databases it 
would be nice for the users to be able to continue doing their stuff while the 
BBS is scanning.

 >  PS> Server #2)
 > Not good over a network - network traffic would be high. Also the coarse
 > grain of integrity - files -would mean frequent contention for data.

I suppose so.

 > Personally, I would love to see/own a BBS server that is written in SQL.
 > I'd write it myself if I had time. The database technology had 30 years
 > of finding out how to do this stuff. Much better than anyone's first
 > attempt at doing it by hand, even yours, I would guess. No insult intended.

None taken. Although maybe you're right, I haven't used SQL. And I really 
wanna get this working by myself. That'll give me pretty good experience in 
this area. And since BBS'es databases aren't THAT big it shouldn't be any 
problem. The only things a bit big is the messagebases on systems saving 
massiges for a long time. That's when the idea of copying the data and then 
scan for stuff goes sour, I message base of 5 megs is slow to copy, and it 
takes up HD, but maybe that's inevitable. I'm still not sure how to 
handle that. Thank you very much for you're opinions and help! 


/ Peter Schuller   [TeamOS/2]   petersch{at}kuai.se


--- AdeptXBBS v1.08a26 (VC) (Registered)