RE: [cp-patches] FYI: Fixlet for DataInputStream.readLine

[Jeroen Frijters]

Hi Roman,

I don't think this is a good idea. It adds a dependency on the internals
of two other classes and assumes invariants that may not even be true
(for example, think of subclasses of BufferedInputStream). It also seems
like it is too complicated for very little gain. Do you have a real
world scenario where performance is hurt by the previous approach?

Regards,
Jeroen

> -----Original Message-----
> From: address@hidden 
> [mailto:address@hidden 
> On Behalf Of Roman Kennke
> Sent: Wednesday, June 15, 2005 17:21
> To: address@hidden
> Subject: [cp-patches] FYI: Fixlet for DataInputStream.readLine
> 
> I added an optimization to DataInputStream.readLine(). If this method 
> needs to push back a character, it used to wrap the field in into a 
> PushbackInputStream. This is not always necessary, especially 
> in cases 
> where in is an InputStream that supports the pos field for 
> repositioning. I committed this.
> 
> 2005-06-15  Roman Kennke  <address@hidden>
> 
>     * java/io/DataInputStream.java
>     (readLine): Added checks for InputStream that support 
> have a pos field
>     for repositioning the stream.
> 
> /Roman
>