# Implementing HTTP/1.1 request chunking **URL:** \ **Category:** NGINX\ **Created:** [January 15, 2010, 8:27am UTC](https://www.ruby-forum.com/t/implementing-http-1-1-request-chunking/181651 "2010-01-15T08:27:39Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![dzo](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/a9adbd/32.png) [@dzo](https://www.ruby-forum.com/u/dzo)\ **Post date:** [January 15, 2010, 8:27am UTC](https://www.ruby-forum.com/t/implementing-http-1-1-request-chunking/181651/1 "2010-01-15T08:27:39Z")
Hi all, I want to get some input with regards to having HTTP/1.1 request chunking in Nginx. I¹ve done some digging, and I¹ve found a few patches and the module developed by agentzh. So, my question is this – What, as a developer, would I need to do so that I could create a patch for this that would be accepted as part of the Nginx distribution? I know that some of the items on this list so far are: 1. That the patch works. 2. Error checking – check the size of the chunks & that the series of chunks are terminated properly. 3. Don’t use Ragel. 🙂 Or, for that matter, any tool not already in use by the Nginx build process. Thank you - Ian H. Greenhoe MANDIANT Senior Software Developer 675 North Washington Street Suite 210 Alexandria, VA 22314 703.224.2905 t 703.683.2891 f 512.422.9564 m removed\_email\_address@domain.invalid > **[Cyber Threat Defense Solutions | Threat Intelligence Services](https://www.mandiant.com/)** > > Mandiant Cyber Threat Defense Solutions leverage innovative technology and expertise from the frontlines to protect your organization against cyber attacks. ---
**Author:** ![dzo](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/a9adbd/32.png) [@dzo](https://www.ruby-forum.com/u/dzo)\ **Post date:** [January 15, 2010, 11:05am UTC](https://www.ruby-forum.com/t/implementing-http-1-1-request-chunking/181651/2 "2010-01-15T11:05:07Z")
On Fri, Jan 15, 2010 at 3:19 PM, Ian G. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > So, my question is this – > What, as a developer, would I need to do so that I could create a patch for > this that would be accepted as part of the Nginx distribution? I started writing the ngx\_chunkin module just because it’s particularly hard (for me) to create a patch (of this size) that can be accepted by Igor S.. Patching the core would be much much easier than doing chunked reading from outside 🙂 > I know that some of the items on this list so far are: > > 1. That the patch works. ngx\_chunkin works 🙂 AFAIK, some people have already put it into production 🙂 > 1. Error checking – check the size of the chunks & that the series of > chunks are terminated properly. ngx\_chunkin honors the http core module’s max\_client\_body\_size directive settings, as well as the client\_body\_buffer\_size and client\_body\_timeout settings. > 1. Don’t use Ragel. 🙂 Or, for that matter, any tool not already in use by > the Nginx build process. Okay, ngx\_chunkin uses Ragel just to eliminate a lot of boring code for truly streaming parsing. Fortunately we can generate C source before we make a release such that the end user don’t need to know anything about Ragel 🙂 I’m not meant to discourage your efforts. If your patch gets accepted by the core, I’d happily mark ngx\_chunkin as obsolete and just encourage people to use it if they have to stick with an earlier nginx version for some reasons. Cheers, -agentzh ---
**Author:** ![dzo](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/a9adbd/32.png) [@dzo](https://www.ruby-forum.com/u/dzo)\ **Post date:** [January 15, 2010, 11:15am UTC](https://www.ruby-forum.com/t/implementing-http-1-1-request-chunking/181651/3 "2010-01-15T11:15:40Z")
On Fri, Jan 15, 2010 at 3:19 PM, Ian G. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > I know that some of the items on this list so far are: Oh, forgot to mention we should also take into account HTTP 1.1 keepalive and pipelining support. ngx\_chunkin currently has a working non-pipelined keepalive support, but the pipelining support is untested so far. (Even though POST + pipelining is rare in real world.) Cheers, -agentzh ---
**Author:** ![dzo](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/a9adbd/32.png) [@dzo](https://www.ruby-forum.com/u/dzo)\ **Post date:** [January 15, 2010, 3:50pm UTC](https://www.ruby-forum.com/t/implementing-http-1-1-request-chunking/181651/4 "2010-01-15T15:50:13Z")
On 1/15/10 5:04 AM, “agentzh” [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > On Fri, Jan 15, 2010 at 3:19 PM, Ian G. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) > wrote: > > > So, my question is this – > > What, as a developer, would I need to do so that I could create a patch for > > this that would be accepted as part of the Nginx distribution? > > I started writing the ngx\_chunkin module just because it’s > particularly hard (for me) to create a patch (of this size) that can > be accepted by Igor S.. Patching the core would be much much > easier than doing chunked reading from outside 🙂 Ooh, yeah. It would definitely be easier to patch the core… Which is why I was asking. > > I know that some of the items on this list so far are: > > > > 1. That the patch works. > > ngx\_chunkin works 🙂 AFAIK, some people have already put it into production 🙂 Didn’t mean to imply that it didn’t – I’m just listing all of the requirements. 🙂 > > 1. Error checking – check the size of the chunks & that the series of > > chunks are terminated properly. > > ngx\_chunkin honors the http core module’s max\_client\_body\_size > directive settings, as well as the client\_body\_buffer\_size and > client\_body\_timeout settings. This is exactly the sort of thing that I need to know. > > 1. Don’t use Ragel. 🙂 Or, for that matter, any tool not already in use by > > the Nginx build process. > > Okay, ngx\_chunkin uses Ragel just to eliminate a lot of boring code > for truly streaming parsing. Fortunately we can generate C source > before we make a release such that the end user don’t need to know > anything about Ragel 🙂 I understand – I’ve both used parser generators and coded parsers by hand. Using parser generators is much preferred. > I’m not meant to discourage your efforts. If your patch gets accepted > by the core, I’d happily mark ngx\_chunkin as obsolete and just > encourage people to use it if they have to stick with an earlier nginx > version for some reasons. Here’s to hoping that I can obsolete your code. 🙂 -Ian