Hi Anders
Anders Brander writes:

> Hummm.... Or something like:
> "... the two domains to be aliased ..." - without saying which is which,
> for the user it doesn't matter much.

Oh Anders, I need rigidly defined areas of doubt and uncertainty!  It's
because I'm a boring old fart that I desperately need to be sure that I'm
doing the right thing when I use something new.  I need to know that
I'm creating an alias domain for an existing domain and not aliasing
an existing domain into the bit bucket.

> A usage like:
> "vaddaliasdomain [options] domain-a.tld domain-b.tld" - nothing to be
> confused about.

But a lot to be scared about unless you KNOW that vaddaliasdomain will
do the right thing [tm] automatically.  I think that the usage has to make
it explicit that whichever way around you put in the arguments, it will do
the right thing.  Look how much effort Larry has to put into explaining
that perl does what you expect (unless you expect something different).

My take on it is that it's far too easy for developers to slip into the
belief that everyone has read and understands the source, or that everyone
has read all the pertinent mailing lists.  It is also my belief that such
an approach is WRONG.  My belief is that software should not, in the
majority of cases, require you to refer to the mailing list.  My belief
is that, in the majority of cases, software should not require you to
read the documentation - it should do what you want it to.

Because vpopmail bridges so many divides, it cannot intuit what you want.
It doesn't know if you're using cdb for everything or using MySQL for
everything or whatever unless you tell it.  But, wherever possible, it
should be DWIM.  Tom's proposed patch allowing real and alias domain in
any order is very much DWIM that pleases both sides of te argument.

Paul Allen
Softflare Support

Reply via email to