Specification vs. Implementation Alex Shinn (23 Aug 2005 02:49 UTC)
Re: Specification vs. Implementation Michael Sperber (23 Aug 2005 07:24 UTC)
Re: Specification vs. Implementation Alex Shinn (24 Aug 2005 02:48 UTC)
Re: Specification vs. Implementation Per Bothner (24 Aug 2005 04:27 UTC)
Re: Specification vs. Implementation Michael Sperber (24 Aug 2005 17:28 UTC)
Re: Specification vs. Implementation Michael Sperber (24 Aug 2005 17:45 UTC)

Re: Specification vs. Implementation Michael Sperber 24 Aug 2005 17:28 UTC

Per Bothner <xxxxxx@bothner.com> writes:

> The current draft says:
>   Streams and ports from the upper layers of the I/O system always
>   perform access through the abstractions provided by [the primitive
>   I/O] layer.
> Hopefully that's not intentional - I certainly don't intend to go
> through the primitive I/O layer to do I/O!

No, it's not.  Thanks for catching it!

> I think the Primitive I/O layer has very limited usefulness.
> It is somewhat similar to JAVA 1.4's 'java.nio' (New I/O) package,
> which I think very few people are using directly, and even fewer
> are using without also using non-blocking I/O.

I agree it's won't be commonly used.  However, for a number of uses
(whenever you create a new kind of data source), it's absolutely
essential.

> Input streams might be useful, but I don't understand what
> output streams are for.

They're there mainly for symmetry, for people who want to live
exclusively in the streams layer.

--
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla