RFR: 6968351: httpserver clashes with delayed TCP ACKs for low Content-Length [v4]

[robert engels]

On Fri, 19 Apr 2024 18:12:14 GMT, Daniel Jeliński <djelinski at openjdk.org> wrote:

>> robert engels has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   fix jtreg params
>
> src/jdk.httpserver/share/classes/sun/net/httpserver/FixedLengthOutputStream.java line 68:
> 
>> 66:         remaining --;
>> 67:         if(remaining==0) {
>> 68:             close();
> 
> `close()` has the side effect of closing the input stream. This again may break other code.

We can remove this. As I stated, it is a bad bug and will leave the connection in a corrupted state if the handler does not call close() on the outstream or the exchange, so I don't think existing code "works".

Also, I think issuing the close() here is benign, because any further writes would throw an exception (too much data), and a subsequent close (or multiple closes) are ignored.

Essentially, once you have written all of the data you stated you would send, it is a critical error if you do anything on the output stream other than flush() or close(). I added a check that a flush() on a closed stream is ignored.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/18667#discussion_r1572757591