lua-users home
lua-l archive

[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index] [Thread Index]

Yes. I know the problem.  Lets ignore the underlying technology
specifics and define the Lua programmers interface. Then
people can tear it down with requirements for things like
overlapped IO and IOCPs. And we can then expand the
interface to accomodate them.
Then we write the code.


On 11/7/06, Wim Couwenberg <> wrote:
Hi David,

> A generic IO interface that could be used for things other than ANSI IO
> I think would be a helpful thing. So even if the implementation is not
> always provided, define the interface. e.g. f:read(), f:write() etc and
> define what to do when a particular IO system does not support the
> interface (e.g. read only systems) and define what to do when the
> target system has more than one IO system.
> This also again raises the question about remove(), rename() being in the
> os library rather than the io library.

One concern is asynchronous i/o.  If we want it, it will cause some
headeaches, since the models differ quite a bit.  (Try to create an
anonymous pipe in Win32 for asynchronous communication...)