Trev Roydhouse wrote in a message to Jack Stein:
 > I guess, but how long does it take to open and close a 
 > file?  Seems pretty darn fast when I open and close 
 > them from any app I've ever written.  
 TR> A handful of files, yes it's as fast as one could want. 
 TR> Unfortunately here we're talking at least a couple of
 TR> hundred.  
I just opened and closed 4000 files in 47 seconds.  I'm figuring with a 
database format, I'd save less than 47 seconds by not opening and closing 
files.
 TR> Then there's the time taken to do a directory scan with
 TR> findfirst()/findnext().  
A database has to impliment the same search on the base, does it not?
 > I'm not even sure how ORENUM does it anyway, but I'll bet 
 > opening and closing files is an insignificant amount of time 
 > in the renumbering process, but I'd be taking a wild 
 > guess there.  
 TR> It is the most significant time (at least in the versiuon of
 TR> renum that I wrote) if you kill messages by date, as every
 TR> single message has to be opened and parsed and closed. 
I'm still thinking reading a database format requires the same parsing of the 
date, less the overhead of opening/closing the file which in my test takes 
47/4000 seconds.
 
 TR> In most other situations, the message file renaming process (ie
 TR> the actual renumbering) takes about the same time as you
 TR> open()/close() each message twice.  
I just checked my logs for this month, and it took 10 minutes to renumber 
100
new messages one day and 10 minutes to renumber 192 messages another day.  I 
keep a low water mark of between 6 and 7000 messages.  I don't know where the 
time is spent, but less than 2 minutes of it should be opening and closing 
files.
 > Rebuilding the index would be the long part, wouldn't it?  
 TR> Have you ever seen how fast my Opus user database utility
 TR> re-indexes :-)  
Sure, all the time.  Just deleted a John Doe login, took about 15 seconds on 
a 1500 user base.  That's why I always smile when your name pops up on any 
OPUS software:-)  (I would like a log of eveyone I ever Zapped though)
About the only OPUS stuff I have that I really dis-like is OEVENT... what a 
nightmare.  I've gone to copying sched.dat from back-up at time changes 
rather than mess with it any more.  It works though, so no, I'm not going to 
write a replacement, even if I could:-)
                                              Jack 
--- timEd/2-B11
---------------