David Daney wrote:
The basic problem was that ChunkedInputStream.skip() would skip on the
underlying InputStream. This bypassed the byte counting of the chunked
protocol resulting in failure on reads following the skip.
The fix is to extent InputStream instead of FilterInputStream, as its
skip method magically does the right thing. This requires that an
implementation of available() and close() also be supplied. We can also
use the super class' implementation of read(byte[]), so I removed that
as well.
I am going to test this patch a bit more, but I think it is generally
the right approach.
Comments?
2006-02-16 David Daney <[EMAIL PROTECTED]>
PR classpath/26312
* gnu/java/net/protocol/http/ChunkedInputStream.java (imports): Cleaned
up.
(ChunkedInputStream): Extend InputStream.
(in): New field.
(headers): Moved to top of class.
(constructor): Save referenct to in.
(read(byte[])): Removed method.
(read(byte[], int, int)): Made synchronized and throw IOException
on error parsing chunk header.
(available): New method.
(close): New method.
After more testing I committed the patch as posted.
2006-02-27 David Daney <[EMAIL PROTECTED]>
PR classpath/26312
* gnu/java/net/protocol/http/ChunkedInputStream.java (imports): Cleaned
up.
(ChunkedInputStream): Extend InputStream.
(in): New field.
(headers): Moved to top of class.
(constructor): Save referenct to in.
(read(byte[])): Removed method.
(read(byte[], int, int)): Made synchronized and throw IOException
on error parsing chunk header.
(available): New method.
(close): New method.
