> This is the dilemma. I realise I'm stating the obvious, in that when
> distributing an application one has to cater for all scenarios, not only

on

> installation but in the future. If computer name works in all cases I'll
> certainly go with that, but what happens when (a) there is no hosts file

or

> (b) the computer has no name, which i think can be the case in a simple
> stand-alone setup.

I've never seen a Windows computer without a name. I'm not sure you can
leave it blank but I've never tried it. The only way to guarantee success is
to use the IP address instead of the server name. You can either use a
routine to read the IP address or do what I do, save it in the program's ini
file for reference when the program starts. Personally, I prefer to make
recommendations to clients to follow certain "rules" when setting up
computers and networks because of the confusion that can occur, but that's
just me. I find it easier to troubleshoot when I know that at least some
rules are followed always.

> It's just that I had a case earlier this month when a client who had just
> changed to my localhost version (since shelved) reported that he was

unable

> to connect even after making the entry in the hosts file. I suggested that
> he re-boot, after which the problem disappeared. I was not on site, so it
> may have been something else. I'm afraid I don't know which Windows

version

> was involved.

This could have been related to something else mucking up the OS. Some rogue
programs tend to confuse Windows enough for things like this to happen but
don't take it for granted that is what happened. As I said, I never tried
changing the IP address of an existing server in the Hosts file and then
trying to access it without rebooting.