Re: RFR 8184285: Buffer sizes of Flow based BodyProcessor API

Tobias, Your mail provoked me look further into the issue of memory usage in the implementation. I now see that there is certainly the possibility for an extreme amount of memory to be used in the pathological case. I filed an issue to track this ( it has more details ), and have a proposed cha

Re: RFR 8184285: Buffer sizes of Flow based BodyProcessor API

Tobias, Thank you for trying this out and giving such a detailed reply. Comments inline ... On 18 Aug 2017, at 18:45, Tobias Thierer wrote: Hi Michael & Chris - apologies for the slow follow-up. I couldn't get my IDE (IntelliJ) to accept a locally built OpenJDK 9 into which I had patched

Re: RFR 8184285: Buffer sizes of Flow based BodyProcessor API

Tobias, Michael, I did a few rounds of iteration on this, that address Tobias’s comments ( modulo the variable buffer sizing, which we agreed to defer for now unless there is a compelling use case ). With this change, there is zero-copy of data. The composite container, List, is relatively ligh

Re: RFR 8184285: Buffer sizes of Flow based BodyProcessor API

On 04/08/2017, 10:22, Tobias Thierer wrote: Hi Michael - thanks for your work! The fact that BufferingProcessor wraps another BodyProcessor loo

Re: RFR 8184285: Buffer sizes of Flow based BodyProcessor API

Hi Tobias, On 04/08/2017, 10:22, Tobias Thierer wrote: Hi Michael - thanks for your work! The fact that BufferingProcessor wraps another BodyPro