I made those changes and re-submitted the patch  (also moving the
unrelated test suite hierarchy stuff elsewhere).  Is this what you had
in mind?

-Josh

On Fri, Feb 8, 2008 at 4:40 AM, Nick Burch <[EMAIL PROTECTED]> wrote:
> On Thu, 7 Feb 2008, Josh Micich wrote:
> > I didn't think closing the input stream twice was the main worry.  My
> > main concern was the possibility of reading the input stream after
> > reaching EOF.  One way to achieve this would be with a stream that
> > returned true for markSupported().  reset() could be called at any time,
> > and the whole stream could be re-read without re-opening.  (Note - the
> > DB2 driver supports this).
>
> That does give me an idea. How about we check for markSupported() before
> we auto-close the input stream. If it is supported, we issue a warning to
> the user that behaviour has changed. If we were to also provide the
> InputStream wrapper that ignores the close, we can tell them the one line
> fix with the warning.
>
> (We can turn the warning off a release or two later on, once everyone who
> needs to has fixed their code)
>
> Does that sound sensible and workable to people?
>
>
> Nick
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to