[RC5] LIFO? Is this true?

gindrup at okway.okstate.edu gindrup at okway.okstate.edu
Wed May 6 09:32:51 EDT 1998


     1.  LIFO.
     2.  Yup, for the clients.  The proxies are FIFO.
     3.  Could be.  You should occasionally allow a client to "run dry".  
     Given the current 6-8 week subspace completion time, that's at least 
     how often a client should be allowed to exhaust its buff-in.
     4.  It's easier to write if the data starts at a constant offset in 
     the buff-in.  FIFO causes the first data element to wander around 
     the file.  In a naive implementation of FIFO, the data continuously 
     moves further and further back in the file with an ever increasing 
     empty segment in the front.  The simplest solution to this problem 
     requires a buff-in that's twice as large and treats the "slots" as 
     elements in a circular list.
        Remember, blocks that are in-process can't be shuffled around 
     else the client cracking that block would become confused (and it 
     prevents you from doing this to it by locking the block down).  LIFO 
     is easier to implement in current filesystems because (a) current 
     filesystems don't allow the start of a file to be moved arbitrarily 
     around in the file and (b) all the "action" occurs at the end of the 
     file -- a well-defined place.
            -- Eric Gindrup ! gindrup at Okway.okstate.edu


______________________________ Reply Separator _________________________________
Subject: [RC5] LIFO?  Is this true? 
Author:  <rc5 at llamas.net> at SMTP
Date:    5/5/98 10:07 PM


LIFO? Is this true?
If so there must be some blocks at the bottom of my buffer which have been 
festering there for weeks.
Why is it designed this way?
regards,  Barry
     
-----Original Message-----
From: Stephen Berg <sberg at gccweb.net> 
To: rc5 at llamas.net <rc5 at llamas.net>
Date: 05 May 1998 00:38
Subject: [RC5] NT GUI and Dialer
     
     
>...[snip]...
>If I'm online and it's frequently updating I end up with blocks in 
>the buffer never being processed since they're LIFO.
     
     
--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net 
rc5-digest subscribers replace rc5 with rc5-digest
     
     


--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list