🚀 go-pugleaf

"Rodolfo Frino - Stargate Software" <starcinema@eftel.com> wrote in message
news:3e7ee719@newsgroups.borland.com...

> However, you mentioned and then posted (on my request)
> an example that is completely irrelevant (and that you
> wrongly thought it was relevant)

It was not irrelvant.  It directly proved statements you questioned
earlier - the VCL could and was altering the process's current directory at
runtime, and I gave two very specific tests that proved that.  The only
change you made to the project was to call GetCurrentDirectory() once (which
even that was not 100% reliable), whereas I was calling it multiple times to
show that it was indeed being altered over and over at runtime.

> since I can overcome any internal change to the current directory
> done within the application via either an *unofficial*  or an
> *official* start up code.

By caching the value once at startup and reusing it instead of retrieving it
over and over, of course.  That's fine.  That was not the point of my sample
project, though.

> Of course, I cannot overcome an external change

There already is a proven and well-known way to overcome that 100% reliably,
but you don't seem willing to use it yourself.  Using Application->ExeName
is always accurate regardless of any startup code used to spawn the process,
regardless of any internal or external changes to the process at startup or
at runtime.


Gambit