Hi all!

I've just kicked out OpenResty 1.7.0.1 RC1 for testing:

    http://openresty.org/download/ngx_openresty-1.7.0.1rc1.tar.gz

Changes since the last (formal) release, 1.5.12.1:

 *   upgraded the Nginx core to 1.7.0.

     *   see the changes here: <http://nginx.org/en/CHANGES>

 *   feature: bundled new Lua library, the lua-resty-lrucache library, which
     is also enabled by default. see
     https://github.com/openresty/lua-resty-lrucache#readme for more
     details.

 *   upgraded LuaJIT to v2.1-20140531.

     *   imported Mike Pall's latest bug fixes and other changes:

         *   Fix frame traversal while searching for error function.

         *   Fix FOLD rule for STRREF of SNEW.

         *   FFI: Fix recording of indexing a struct pointer ctype
             object itself.

         *   FFI: Another fix for cdata equality comparisons.

         *   Fix FOLD rule for "string.sub(s, ...) == k".

         *   x86: Fix code generation for unused result of
             "math.random()".

         *   x64: Workaround for MSVC build issue.

         *   PPC: Fix red zone overflow in machine code generation.

         *   Fix compatibility issues with Illumos. Thanks to Theo
             Schlossnagle.

         *   Add PS Vita port. Thanks to Anton Stenmark.

     *   disabled trace stitching by default for now since it may
         trigger random lua stack corruptions when using with
         ngx_lua.

     *   feature: jit.dump: output Lua source location after every
         BC.

     *   feature: added internal memory-buffer-based trace
         entry/exit/start-recording event logging, mainly for
         debugging bugs in the JIT compiler. it requires
         "-DLUA_USE_TRACE_LOGS" when building.

     *   feature: save "g->jit_base" to "g->saved_jit_base" before
         "lj_err_throw" clears "g->jit_base" which makes it
         impossible to get Lua backtrace in such states.

 *   upgraded the lua-resty-core library to 0.0.7.

     *   feature: implemented ngx.req.set_header() (partial:
         table-typed values not yet supported) and
         ngx.req.clear_header() with FFI in the resty.core.request
         module.

     *   feature: implemented shdict:flush_all() with FFI in the
         resty.core.shdict.

     *   feature: implemented ngx.req.set_method() with FFI in
         resty.core.request.

     *   feature: implemented ngx.req.get_method() with FFI in
         resty.core.request.

     *   feature: implemented ngx.time() with FFI in resty.core.time.

     *   feature: implemented ngx.req.start_time with FFI in
         rest.core.request.

     *   feature: implemented ngx.now() with FFI in resty.core.time.

 *   upgraded the ngx_lua module to 0.9.8.

     *   bugfix: the ngx.ctx table might be released prematurely when
         ngx.exit() was used to generate the response header. thanks
         Monkey Zhang for the report. now we always release ngx.ctx
         in our request pool cleanup handler.

     *   bugfix: we did not call our coroutine cleanup handlers right
         after our coroutine completes (either successfully or
         unsuccessfully) otherwise segmentation fault might happen
         when the Lua VM throws out unexpected exceptions like
         "attempt to yield across C-call boundary". thanks Lipin
         Dmitriy for the report.

     *   bugfix: nginx does not guarentee the parent pointer of the
         rbtree root is meaningful, which could lead to inifinite
         loops when the ngx_lua module tried to abort pending timers
         prematurely (upon worker exit). thanks pengqi for the
         report.

     *   bugfix: ngx.req.set_method(): we incorrectly modified
         "r->method" when the method ID was wrong.

     *   bugfix: rewrite_by_lua* and access_by_lua* will now
         terminate the current request if the response header has
         already been sent (via calls like ngx.say and
         ngx.send_headers) at that point. thanks yaronli and Sophos
         for the report.

     *   bugfix: issues in the error handling for pure C API
         functions for shared dict. thanks Xiaochen Wang.

     *   feature: now we save the original pattern string pointer
         value into our "ngx_http_lua_regex_t" C struct, to help
         runtime regex profiling and debugging.

     *   feature: allow use of 3rd-party pcre bindings in
         init_by_lua*. thanks ikokostya for the feature request.

     *   feature: added pure C API functions to support the new
         FFI-based Lua API implemented in the lua-resty-core library.

     *   feature: make use of the new shm API in nginx 1.5.13+ to
         suppress the "no memory" error logging when the shared
         dictionaries run out of memory.

     *   feature: added C macro "NGX_LUA_ABORT_AT_PANIC" to allow
         generating a core dump when the Lua VM panics.

 *   upgraded the ngx_srcache module to 0.27.

     *   bugfix: we used to skip all the output header and body
         filters run before our filters (which unfortunately bypassed
         the standard ngx_http_not_modified_filter_module, for
         example). thanks Lloyd Zhou for the report.

     *   feature: added new config directive srcache_store_ranges for
         storing 206 Partial Content responses generated by the
         standard ngx_http_range_filter_module.

 *   bugfix: updated the dtrace patch because systemtap 2.5 no longer
     accepts the "-xnolib" option in its dtrace utility.

 *   removed our bundled version of "ngx_http_auth_request_module"
     because recent versions of the nginx core already has it. thanks
     LazyZhu for the report.

 *   bugfix: applied our patch for the nginx core to fix the long
     standing memory fragmentation issue for blocks larger than the
     page size in the nginx slab allocator:
     <http://mailman.nginx.org/pipermail/nginx-devel/2014-May/005316.html>

You're very welcome to try out this RC release on your side. Any
feedback will be highly appreciated. Thank you!

Best regards,
-agentzh
    双工socket还没出来。。。

    在 2014年6月1日星期日UTC+8下午2时25分38秒,agentzh写道:
    Hi all!

    I've just kicked out OpenResty 1.7.0.1 RC1 for testing:

        http://openresty.org/download/ngx_openresty-1.7.0.1rc1.tar.gz

    Changes since the last (formal) release, 1.5.12.1:

     *   upgraded the Nginx core to 1.7.0.

         *   see the changes here: <http://nginx.org/en/CHANGES>

     *   feature: bundled new Lua library, the lua-resty-lrucache library, which
         is also enabled by default. see
         https://github.com/openresty/lua-resty-lrucache#readme for more
         details.

     *   upgraded LuaJIT to v2.1-20140531.

         *   imported Mike Pall's latest bug fixes and other changes:

             *   Fix frame traversal while searching for error function.

             *   Fix FOLD rule for STRREF of SNEW.

             *   FFI: Fix recording of indexing a struct pointer ctype
                 object itself.

             *   FFI: Another fix for cdata equality comparisons.

             *   Fix FOLD rule for "string.sub(s, ...) == k".

             *   x86: Fix code generation for unused result of
                 "math.random()".

             *   x64: Workaround for MSVC build issue.

             *   PPC: Fix red zone overflow in machine code generation.

             *   Fix compatibility issues with Illumos. Thanks to Theo
                 Schlossnagle.

             *   Add PS Vita port. Thanks to Anton Stenmark.

         *   disabled trace stitching by default for now since it may
             trigger random lua stack corruptions when using with
             ngx_lua.

         *   feature: jit.dump: output Lua source location after every
             BC.

         *   feature: added internal memory-buffer-based trace
             entry/exit/start-recording event logging, mainly for
             debugging bugs in the JIT compiler. it requires
             "-DLUA_USE_TRACE_LOGS" when building.

         *   feature: save "g->jit_base" to "g->saved_jit_base" before
             "lj_err_throw" clears "g->jit_base" which makes it
             impossible to get Lua backtrace in such states.

     *   upgraded the lua-resty-core library to 0.0.7.

         *   feature: implemented ngx.req.set_header() (partial:
             table-typed values not yet supported) and
             ngx.req.clear_header() with FFI in the resty.core.request
             module.

         *   feature: implemented shdict:flush_all() with FFI in the
             resty.core.shdict.

         *   feature: implemented ngx.req.set_method() with FFI in
             resty.core.request.

         *   feature: implemented ngx.req.get_method() with FFI in
             resty.core.request.

         *   feature: implemented ngx.time() with FFI in resty.core.time.

         *   feature: implemented ngx.req.start_time with FFI in
             rest.core.request.

         *   feature: implemented ngx.now() with FFI in resty.core.time.

     *   upgraded the ngx_lua module to 0.9.8.

         *   bugfix: the ngx.ctx table might be released prematurely when
             ngx.exit() was used to generate the response header. thanks
             Monkey Zhang for the report. now we always release ngx.ctx
             in our request pool cleanup handler.

         *   bugfix: we did not call our coroutine cleanup handlers right
             after our coroutine completes (either successfully or
             unsuccessfully) otherwise segmentation fault might happen
             when the Lua VM throws out unexpected exceptions like
             "attempt to yield across C-call boundary". thanks Lipin
             Dmitriy for the report.

         *   bugfix: nginx does not guarentee the parent pointer of the
             rbtree root is meaningful, which could lead to inifinite
             loops when the ngx_lua module tried to abort pending timers
             prematurely (upon worker exit). thanks pengqi for the
             report.

         *   bugfix: ngx.req.set_method(): we incorrectly modified
             "r->method" when the method ID was wrong.

         *   bugfix: rewrite_by_lua* and access_by_lua* will now
             terminate the current request if the response header has
             already been sent (via calls like ngx.say and
             ngx.send_headers) at that point. thanks yaronli and Sophos
             for the report.

         *   bugfix: issues in the error handling for pure C API
             functions for shared dict. thanks Xiaochen Wang.

         *   feature: now we save the original pattern string pointer
             value into our "ngx_http_lua_regex_t" C struct, to help
             runtime regex profiling and debugging.

         *   feature: allow use of 3rd-party pcre bindings in
             init_by_lua*. thanks ikokostya for the feature request.

         *   feature: added pure C API functions to support the new
             FFI-based Lua API implemented in the lua-resty-core library.

         *   feature: make use of the new shm API in nginx 1.5.13+ to
             suppress the "no memory" error logging when the shared
             dictionaries run out of memory.

         *   feature: added C macro "NGX_LUA_ABORT_AT_PANIC" to allow
             generating a core dump when the Lua VM panics.

     *   upgraded the ngx_srcache module to 0.27.

         *   bugfix: we used to skip all the output header and body
             filters run before our filters (which unfortunately bypassed
             the standard ngx_http_not_modified_filter_module, for
             example). thanks Lloyd Zhou for the report.

         *   feature: added new config directive srcache_store_ranges for
             storing 206 Partial Content responses generated by the
             standard ngx_http_range_filter_module.

     *   bugfix: updated the dtrace patch because systemtap 2.5 no longer
         accepts the "-xnolib" option in its dtrace utility.

     *   removed our bundled version of "ngx_http_auth_request_module"
         because recent versions of the nginx core already has it. thanks
         LazyZhu for the report.

     *   bugfix: applied our patch for the nginx core to fix the long
         standing memory fragmentation issue for blocks larger than the
         page size in the nginx slab allocator:
         <http://mailman.nginx.org/pipermail/nginx-devel/2014-May/005316.html>

    You're very welcome to try out this RC release on your side. Any
    feedback will be highly appreciated. Thank you!

    Best regards,
    -agentzh
      Have you tested with 1.7.1?  It has some important bug fixes and features (syslog for logs in particular). Need any help on that?
        Hi Brian!
        
        On Mon, Jun 2, 2014 at 5:59 AM, Brian Akins wrote:
        > Have you tested with 1.7.1?  It has some important bug fixes and features
        > (syslog for logs in particular). Need any help on that?
        >
        
        I need to get OpenResty 1.7.0.1 out first :) And then I'll upgrade to
        NGINX 1.7.1 or better (if available) as soon as possible. Testing a
        new release of NGINX on our side requires some time.
        
        Best regards,
        -agentzh
        
          I understand about a new release taking a bit of time.  Let me know how I can help.  
            Hi Brian!
            
            2014-06-02 13:07 GMT-07:00 Brian Akins:
            > I understand about a new release taking a bit of time.  Let me know how I
            > can help.
            >
            
            When the corresponding RC releases are out, it'll be great if you can
            test them on your side. Testing the current RC release on your side
            would also be helpful :)
            
            Thanks!
            -agentzh
            
              Write a Reply...