Re: another NULL/reference hole in the API

[Daniel Elstner <daniel kitta googlemail com>]

Am Dienstag, den 14.08.2007, 12:09 -0400 schrieb Paul Davis:

> I've just discovered that all the spawn functions in glibmm make it
> impossible to pass NULL to the underlying C layer to indicate "use same
> working directory as parent". This is another instance of a general
> pattern/hole that has been mentioned recently. I don't recall what
> Murray's position on it is. Do we have a general policy?

For strings we usually interpret the empty string as NULL, since in most
cases you wouldn't ever want to pass an actual empty string.  For the
few functions where you do want that we introduce overloads.

--Daniel