On Wed, 28 Jan 2026 16:53:59 GMT, Artur Barashev <[email protected]> wrote:

>> Implement certificate compression in TLS 1.3 using internally supported ZLIB 
>> compression algorithm. See RFC 8879 for more details:
>> https://datatracker.ietf.org/doc/html/rfc8879
>
> Artur Barashev has updated the pull request incrementally with one additional 
> commit since the last revision:
> 
>   Force cache limit per compression algorithm

> _Mailing list message from [Xuelei Fan](mailto:[email protected]) on 
> [security-dev](mailto:[email protected]):_
> 
> On Tue, Jan 27, 2026 at 9:27?AM Artur Barashev <abarashev at openjdk.org> 
> wrote:
> 
> > On Tue, 27 Jan 2026 16:54:28 GMT, Xue-Lei Andrew Fan <xuelei at openjdk.org>
> > wrote:
> > > > Artur Barashev has updated the pull request incrementally with one
> > > > additional commit since the last revision:
> > > > Allocate 24 bits for input size in cache key. Add unit tests.
> > > 
> > > 
> > > src/java.base/share/classes/javax/net/ssl/SSLParameters.java line 982:
> > > > 980:             boolean enableCertificateCompression) {
> > > > 981:         this.enableCertificateCompression =
> > > > enableCertificateCompression;
> > > > 982:     }
> > > 
> > > 
> > > Is there a plan to support brotli compression algorithm in OpenJDK?  It
> > > is the only supported algorithm in browser Chrome.
> > > If there is a need to support more than one compression algorithms in
> > > the future, it might be better to provide an option to customize the
> > > algorithms selection, including preferences.  The flexibility could 
> > > provide
> > > better interoperability if a vendor does not support compression algorithm
> > > properly.
> > 
> > 
> > Yes, strangely Chrome supports only brotli although I didn't see any
> > difference in compression ration between zlib and brotli when manually
> > testing certificate compression. We may add support for more algorithms in
> > the future, but for now it's zlib only.
> 
> Once there is a need to support more algorithms, this API may not be 
> applicable for some cases especially when there are compatibility impact. It 
> might be good to consider the API to support multiple algorithms from day 
> zero.
> 
> Just my $0.02.
> 
> Thanks, Xuelei
> 
> > Generally speaking, it's the certificate's metadata that is being
> > compressed, the keys are high-entropy data difficult to compress. PQC
> > certificates don?t compress well, most likely because of lower metadata to
> > keys ratio.
> 
> -------------- next part -------------- An HTML attachment was scrubbed... 
> URL: 
> <https://mail.openjdk.org/pipermail/security-dev/attachments/20260128/7e41c352/attachment.htm>

I know original design had callback functions for custom deflators/inflators, 
but it was decided to support only internally implemented ZLIB for now. We may 
add callbacks at the later time if there is a demand for it. Current API only 
enables/disables the certificate compression overall, there shouldn't be any 
compatibly issues with future callback function mechanism if we decide to 
support it.

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

PR Comment: https://git.openjdk.org/jdk/pull/28682#issuecomment-3812904883

Reply via email to