mark lewis wrote in a message to Roy J. Tellason:

 ml>> why would you /have/ to renumber? the DOS 8.3 system can
 ml>> count from 0 thru 99999999 before wrapping... that's a lot
 ml>> of messages :wink: imagine that on a system that handles
 ml>> long file names... them numbers would get pretty large
 ml>> before a renumber was required...

 RJT> Because with *.MSG being stored one message-per-file,  the
 RJT> software (at least some of it) would look for each and every
 RJT> number,  starting from 1,  by way of dos filesystem calls,
 RJT> which was very slow and inefficient.  Putting them all
 RJT> together with consecutive low numbers does away with this
 RJT> problem.

 ml> that's outrageous! what's a highwater mark or lastread pointer in
 ml> the userfile for??

To keep track of "lastread",  which isn't the same thing as
finding out what's there...?

I dunno,  I ran *.msg for a short while when I first started to run a point
system,  and switched to Squish bases as soon as I could manage it.  I'm
not 100% sure,  but I think it was George Peace that told me that.  Nor do
I remember what software it was that did that,  I believe he was running
Opus at the time.

 RJT> OTOH so would a linked list...

 ml>> FWIW: i have _not_ renumbered my message areas in almost a
 ml>> decade...

 RJT> And you're using *.msg?

 ml> i've a couple of them... the rest are all JAM bases... none have
 ml> been renumbered in at least that long...

I've never used those,  and have no idea as to how they would act.

---