poi-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chris Nokleberg <ch...@sixlegs.com>
Subject Re: Slow InputStream patch
Date Sun, 22 Jun 2003 19:18:55 GMT
Andrew C. Oliver wrote:
> Okay, I guess I still donĀ¹t understand why this is necessary given a
> BufferedInputStream is blocking..?

This is covered in the bug report, but the short answer is that
BufferedInputStream is not blocking. It does loop, but will not read
from the underlying stream if available() returns 0. However, even if it
did block, it should not be up to the user to pass wrapped streams to
POI. This patch makes the POI code obey the InputStream contracts, which
is where I feel the responsibility should lie.

> Secondly, does this affect performance in any negative manner?

No, it's negligible.

> If this is non-harmful and necessary then fine.  I'm still just a bit
> confused.

No problem.


View raw message