MB> That has not been a problem as there are only a few 
 MB> thousand messages there, but the message numbers are 
 MB> getting up there (in the 22,000 range).

 BJ> Ok, then why haven't you picked up an external renumber routine for
 BJ> your use and avoid the possible problems?  

 MB> I'd love to, but do not know which are still OK (I've 
 MB> heard that there are Y2K issues with CONFMAIL, which I 
 MB> used previously, for example).  Do you have a 
 MB> recommendation?  It would be appreciated.

Ah....  Part of the reason I've switched from *.msg to squish for almost
all my areas.  I have some limited netmail areas that are *.msg, but they
never get more than a few hunderd messages in a given area before I archive
stuff and start the counters overs.....  So, it has been years since I've
used a renumber function.  

On the other hand, if I needed to do it today, I'd make a copy of the area
(as a backup) and then run sqconv on the area, converting from *.msg to
*.msg, in a new subdirectory.  If that didn't work, I'd run the program
twice, first to go *.msg to squish and the second to go squish to
*.msg.....  I believe that would do the trick....  But, as you know, I'd
just convert most areas to squish and leave it that way....   Just make
sure nothing else is touching the message area you are working on while
performing the conversion.....

 BJ> My guess is that things should run fine till the 32K mark, and
 BJ> might run fine to a higher number....  If you don't want to use a
 BJ> renumber routine, then you need to look at the code....  

 MB> It is not that I don't *want* to run a renumber 
 MB> routine, I just don't *know* what is safe to use now.

Ok...  I seem to recall "renum" was around.  Last time I used it
was prior to Y2K thou....  As long as you are just renumbering the messages
(back to 1.msg, 2.msg, etc.) and not deleting based on time, I would think
the code should be fine....  Make a backup copy and test.....

 BJ>   Basead on
 BJ> your comments (SQSH_111), I suspect you are running the 16 bit DOS
 BJ> version, not a 32 bit version.  The 16 bit version has lower limits
 BJ> on what it can handle. 

 MB> That is what I was wary of, and why the question.

Ok.  I would expect you to be ok to the 32K mark for the numbers.  Actual
number of messages is probably limited to 15K or 16K for indexes stored in
memory (and maybe less if you have less memory).  Since you are only
keeping a few thousand messages, the 15K or 16K assumed index limit won't
be hit.  And it could be good to 64K for the message number, or even to
99999999 if the message number is totally handled as character strings.

Basically, at this point....  make a back up copy and then test the renum
code you were using with out deleting any messages based on date / time
considerations.

Good luck.....

Take care.....

Bob Jones, 1:343/41


--- Maximus/2 3.01