Re: HTTP 2 and Apache HTTP client
On Fri, 2015-01-09 at 10:20 -0700, Shawn Heisey wrote: On 1/9/2015 2:41 AM, Oleg Kalnichevski wrote: Shawn, I can work on HC in my spare time only. Even if I quit my day job and divorce my wife I am not sure HTTP/2.0 in HC by the end of February is realistic. The best case scenario would be having BETA quality support for HTTP/2.0 by the end of year. Oleg I would never make any actual demands of your time, I'm just conveying my opinion that I think it's really important. Thank you for everything you have done and anything you may do in the future. If there is any way my limited http skillset and nonexistent understanding of the codebase can be helpful, I am willing. Other people are likely to be more insistent and demanding once the spec is published. :) Shawn Hi Shawn There are enough folks involved in this project who contribute to development of higher level components such as the caching module or specific auth schemes without getting exposed too much to the lower level transport components. My involvement in this project started with development of cookie specs until I got sucked into development of low level stuff. If you are interested, willing and have some spare bandwidth there will always be something you could contribute to depending on your individual inclinations. Cheers Oleg - To unsubscribe, e-mail: httpclient-users-unsubscr...@hc.apache.org For additional commands, e-mail: httpclient-users-h...@hc.apache.org
Re: HTTP 2 and Apache HTTP client
On Thu, 2015-01-08 at 23:15 -0700, Shawn Heisey wrote: On 1/8/2015 5:28 PM, Stefan Magnus Landrø wrote: Maybe consider jetty instead? https://github.com/eclipse/jetty.project/blob/jetty-http2/jetty-http2/http2-client/src/test/java/org/eclipse/jetty/http2/client/Client.java For an Apache project like Solr (which uses HttpClient and is my primary reason for being here), I see two problems with that idea: One problem is the eating your own dog food argument -- if software from an Apache project fits all of the requirements and has no significant disadvantages compared to a competitor, we should choose to use the software from Apache. Both projects are likely to benefit. http://en.wikipedia.org/wiki/Eating_your_own_dog_food The other problem is that switching to a completely different http client would likely involve significant development time, both for the switch itself and for fixing the inevitable bugs that are a direct result. Seeing that competing projects are already focusing effort on HTTP/2 is even more reason to accelerate HttpClient towards HTTP/2. Even Oracle has fairly concrete plans for a new client in Java itself: https://bugs.openjdk.java.net/browse/JDK-8042950 Thanks, Shawn Shawn, I can work on HC in my spare time only. Even if I quit my day job and divorce my wife I am not sure HTTP/2.0 in HC by the end of February is realistic. The best case scenario would be having BETA quality support for HTTP/2.0 by the end of year. Oleg - To unsubscribe, e-mail: httpclient-users-unsubscr...@hc.apache.org For additional commands, e-mail: httpclient-users-h...@hc.apache.org
Re: HTTP 2 and Apache HTTP client
On 1/9/2015 2:41 AM, Oleg Kalnichevski wrote: Shawn, I can work on HC in my spare time only. Even if I quit my day job and divorce my wife I am not sure HTTP/2.0 in HC by the end of February is realistic. The best case scenario would be having BETA quality support for HTTP/2.0 by the end of year. Oleg I would never make any actual demands of your time, I'm just conveying my opinion that I think it's really important. Thank you for everything you have done and anything you may do in the future. If there is any way my limited http skillset and nonexistent understanding of the codebase can be helpful, I am willing. Other people are likely to be more insistent and demanding once the spec is published. :) Shawn - To unsubscribe, e-mail: httpclient-users-unsubscr...@hc.apache.org For additional commands, e-mail: httpclient-users-h...@hc.apache.org
Re: HTTP 2 and Apache HTTP client
Maybe consider jetty instead? https://github.com/eclipse/jetty.project/blob/jetty-http2/jetty-http2/http2-client/src/test/java/org/eclipse/jetty/http2/client/Client.java Sendt fra min iPhone Den 8. jan. 2015 kl. 23.28 skrev Shawn Heisey apa...@elyograg.org: On 1/8/2015 7:20 AM, Oleg Kalnichevski wrote: There is no concrete roadmap for HTTP/2.0 yet. An immediate objective is full compliance with the latest HTTP/1.1 spec (RFC 7230 and related) https://wiki.apache.org/HttpComponents/HttpComponentsRoadmap Recently I read through the draft RFC for HTTP/2.0. Once everybody has stable and optimized implementations, the average Internet experience (and a lot of behind-the-scenes infrastructure that use HTTP) will improve quite a lot. It looks really awesome. It will be a major development effort to reach a stable 2.0 implementation. The current schedule says that the final RFC will be published in February 2015, which is quite a lot sooner than I thought. http://en.wikipedia.org/wiki/HTTP/2#Development_milestones IMHO, work on the implementation should be underway already, so that there is an initial implementation available by the time the RFC is published. That's easy for me to say, since I am not qualified for the work. If I had any idea how to write code that functions at such a low level, I would have already sent some ideas to Oleg! Thanks, Shawn - To unsubscribe, e-mail: httpclient-users-unsubscr...@hc.apache.org For additional commands, e-mail: httpclient-users-h...@hc.apache.org
Re: HTTP 2 and Apache HTTP client
On 1/8/2015 5:28 PM, Stefan Magnus Landrø wrote: Maybe consider jetty instead? https://github.com/eclipse/jetty.project/blob/jetty-http2/jetty-http2/http2-client/src/test/java/org/eclipse/jetty/http2/client/Client.java For an Apache project like Solr (which uses HttpClient and is my primary reason for being here), I see two problems with that idea: One problem is the eating your own dog food argument -- if software from an Apache project fits all of the requirements and has no significant disadvantages compared to a competitor, we should choose to use the software from Apache. Both projects are likely to benefit. http://en.wikipedia.org/wiki/Eating_your_own_dog_food The other problem is that switching to a completely different http client would likely involve significant development time, both for the switch itself and for fixing the inevitable bugs that are a direct result. Seeing that competing projects are already focusing effort on HTTP/2 is even more reason to accelerate HttpClient towards HTTP/2. Even Oracle has fairly concrete plans for a new client in Java itself: https://bugs.openjdk.java.net/browse/JDK-8042950 Thanks, Shawn - To unsubscribe, e-mail: httpclient-users-unsubscr...@hc.apache.org For additional commands, e-mail: httpclient-users-h...@hc.apache.org
Re: HTTP 2 and Apache HTTP client
On 1/8/2015 7:20 AM, Oleg Kalnichevski wrote: There is no concrete roadmap for HTTP/2.0 yet. An immediate objective is full compliance with the latest HTTP/1.1 spec (RFC 7230 and related) https://wiki.apache.org/HttpComponents/HttpComponentsRoadmap Recently I read through the draft RFC for HTTP/2.0. Once everybody has stable and optimized implementations, the average Internet experience (and a lot of behind-the-scenes infrastructure that use HTTP) will improve quite a lot. It looks really awesome. It will be a major development effort to reach a stable 2.0 implementation. The current schedule says that the final RFC will be published in February 2015, which is quite a lot sooner than I thought. http://en.wikipedia.org/wiki/HTTP/2#Development_milestones IMHO, work on the implementation should be underway already, so that there is an initial implementation available by the time the RFC is published. That's easy for me to say, since I am not qualified for the work. If I had any idea how to write code that functions at such a low level, I would have already sent some ideas to Oleg! Thanks, Shawn - To unsubscribe, e-mail: httpclient-users-unsubscr...@hc.apache.org For additional commands, e-mail: httpclient-users-h...@hc.apache.org