[Dxspider-support] Dx Spider & ARC same computer.

[Dirk Koopman]

On Wed, 2005-01-12 at 01:07 -0800, Lee Sawkins wrote:
> For anyone who is interested, I am running Dx Spider (VE7CC-1) and AR
> Cluster (VE7CC) in the same computer.  Win 2000 Pro.  Athlon 1600+ with
> 256 MB of RAM.  High speed cable internet.  

> I wrote an application that goes between a cluster and the Internet.  It
> completely prevents the "you are already connected" problem.  When it
> detects the same call trying to reconnect, it will disconnect the old
> link.  This application allows remote monitoring of any user connecton
> to the cluster.  It also allows locking out of specific IP addresses.  I
> am running two instances of this program in the computer.
> 

I shall add a flag that allows sysops to choose whether they will allow
the bumping off of existing instances of a callsign.

> I am also running another program I wrote called AR User.  It connects
> to my AR Cluster and once per hour scans the message data base and
> compares unread mail to the current users.  If it finds a match, it
> sends them a talk messages telling them to read their mail.  It also
> detects incoming Dx Spots where the spotted dx is a current user.  It
> will send that user a talk message with the dx spot in it. This is in
> case the user has his spot filters set to reject spots from the area
> where he was spotted from.  As soon as I have time I will modify this
> program to do the same with Dx Spider users.

There is no particular reason why I can't scan the messages and do the
same thing. It is easy enough. There were some reasons, years ago, where
this might not have been ideal, but I was never really convinced.

> 
> Running both cluster applications in the same machine allows direct
> speed comparison between them.  Dx Spider seems to be at least 20 times
> faster looking up Dx spots.  Typically 0.1 seconds vs over 2 seconds. 
> Better filtering too!

This, I find, is really strange. I am a known M$-a-phobe, but unless
someone has the DB indexed wrong / non-existantly, even on a M$ piece of
software, I cannot see how a linear search (which is what sh/dx does)
can be slower than a indexed lookup. This does not make any sense to me.

Also, perl is an interpreted language. DXSpider is completely
interpreted at runtime. There is no compilation. As I understood it AR-C
is compiled VB. ???

This is all slightly worrying as I am considering, optionally, storing
stuff in databases (anything that DBI supports, which is more or less
*any* database [including M$ via ADO or ODBC]). I can see I am going to
have to do some fairly careful benchmarking...

Dirk G1TLH