On Mon, Jul 07, 2003 at 10:37:36AM +0200, Lars Gullik Bjønnes wrote:
> | lyx::InsetCode would do for me personally.
>
> with lyx::insets::Code you could use just Code in a lot of places.
I know.
But even if I am a big fan of short names in the short range, I find them
confusing on a long distance, so long range names better be descriptive.
In fact this lyx::insets::Code illustrates nicely the cognitive problem
of deeply nested namespaces:
We'd basically end up with two new words to learn: short distance 'Code'
and long distance 'inset::Code'. Moreover, most of the uses will be just
'Code', so _I_ lose the connection between 'Code' and 'inset'. When I need
it from a long distance, I don't remember the connection or I have actively
think about it. This slows me down.
I am also a bit wary as I did exactly the same thing about three years
ago (I could check CVS for specific dates) in my day time job project - and
undid it shortly afterwards. So this was a complete waste of time.
A specific reason (but not the only one) was that there is no way to put
a single function inside another namespace. So the code was something like
namespace ge {
namespace comp {
namespace algo {
struct index {};
}
}
namespace io {
void out(ge::comp::index const &) {...}
void in(ge::comp::index &) {...}
}
namespace comp {
namespace algo {
void implementation_using_index_and_io() {..>}
}
}
}
So this is 20 lines just to declare three functions and a struct. No good.
[I think I would try again if the language allowed me to write
namespace ge::comp::algo {
struct index {};
void ::ge::io::out(index const &) {...}
void ::ge::io::in(index &) {...}
void implementation_using_index_and_io() {..>}
}
but it doesn't]
Andre'
--
Those who desire to give up Freedom in order to gain Security, will not have,
nor do they deserve, either one. (T. Jefferson or B. Franklin or both...)
