Oleg, I did as you suggested and modified TestChunkCoding.java to force the situation. The test ran correctly. Sorry for the noise, but thanks for pointing me to the tests.
Dan On Jul 1, 2011, at 3:23 PM, Oleg Kalnichevski wrote: > On Fri, 2011-07-01 at 14:43 -0400, dan evans wrote: >> httpcore-4.1.1 - In ChunkedInputStream, the CharArrayBuffer allocated to >> read chunk metadata has a length of 16. In looking at >> AbstractSessionInputBuffer, it seems that if the chunk has an optional >> "chunk extension" that exceeds this length (i.e. locateLF() returns -1), >> either the stream will get a false EOF or the first part of the metadata >> will be overwritten. This is just a hypothesis from staring at the code. I >> have never seen chunk metadata this long, nor have I been able to actually >> test this. >> >> dan >> > > Dan > > CharArrayBuffer should get automatically expanded in order to fit in a > complete line. There should be a test case for that in the HttpCore test > suite > > http://svn.apache.org/repos/asf/httpcomponents/httpcore/trunk/httpcore/src/test/java/org/apache/http/impl/TestSessionBuffers.java > > However, feel free to add a test case specifically for the chunk codec > > http://svn.apache.org/repos/asf/httpcomponents/httpcore/trunk/httpcore/src/test/java/org/apache/http/impl/io/TestChunkCoding.java > > Oleg > > > > --------------------------------------------------------------------- > 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]
