> Amos Jeffries wrote: >>> I'm working through a problem where pages from a broken web server >>> aren't making it through Squid 3.0 (tested up through STABLE 8). The >>> page source according to the browser (Firefox 2, Firefox 3, or IE7) is >>> the "28B" which should be the start of the chunk, followed by the chunk >>> itself, and then "0" that would finish the chunked transfer. >>> I'm hoping that I can manage to contact the admins of the web server >>> and >>> get them to fix their problem, but is there any way I can work around >>> the problem in Squid? >>> >> >> Squid 3.0 does not support chunks. There is a config workaround to drop >> the Accept-Encoding header on requests going to known broken HTTP/1.1 >> servers which try to send chunks to 1.0 clients. >> >> http://squidproxy.wordpress.com/2008/04/29/chunked-decoding/ >> >> If you really need chunked and 3.x you will need the HEAD code. >> >> Amos >> > Thanks, Amos. How is the stability of the HEAD code, and is 3.1 nearing > release? I believe HEAD/3.1 is ready for use. It's been relatively stable for a while now with no new major problems found. Just two holdups on the formal release: 1) A few bugs to zap and prove dead. We don't release with known bugs rated critical or blocker (just 2 left! and patches ready for those). 2) Final confirmation that AsyncCalls feature is stable. And documented by those who wrote it. This documenting is needed to aid good debugging and proper fixes in the new design once its widely used. They seem quite busy right now so the formal package may be further delayed. But the daily snapshots are IMO stable enough for beta testing or use now by anyone who needs to. Amos
