[njs] Fixed building with GCC ASan.

details: http://hg.nginx.org/njs/rev/d8b1e2576409 branches: changeset: 246:d8b1e2576409 user: Valentin Bartenev date: Thu Nov 10 15:33:02 2016 +0300 description: Fixed building with GCC ASan. diffstat: nxt/auto/memalign | 9 - 1 files changed, 8 insertions(+), 1 deletions

[njs] Fixed building on Solaris after d8b1e2576409.

details: http://hg.nginx.org/njs/rev/9f6d26f5f99b branches: changeset: 247:9f6d26f5f99b user: Valentin Bartenev date: Thu Nov 10 16:19:13 2016 +0300 description: Fixed building on Solaris after d8b1e2576409. diffstat: nxt/auto/memalign | 2 +- 1 files changed, 1 insertions(+), 1

[njs] Removed misused __attribute__((malloc)).

details: http://hg.nginx.org/njs/rev/60c2930eb951 branches: changeset: 248:60c2930eb951 user: Valentin Bartenev date: Thu Nov 10 16:47:52 2016 +0300 description: Removed misused __attribute__((malloc)). According to the documentation: | This tells the compiler that a function is ma

Re: [PATCH] proxy-protocol dst variables and proxy-proxy-protocol

Hello! On Thu, Nov 10, 2016 at 01:06:54AM +0100, Bjørnar Ness wrote: [...] > > Current question is: > > > > What "listen ... proxy_protocol" should mean in case of mail. In > > other modules, it just means that PROXY protocol header is parsed > > and appropriate variables are available for use.

[njs] Improved UTF-8 offset map related macros (no functional ch...

details: http://hg.nginx.org/njs/rev/187882f1895a branches: changeset: 249:187882f1895a user: Valentin Bartenev date: Thu Nov 10 18:45:10 2016 +0300 description: Improved UTF-8 offset map related macros (no functional changes). diffstat: njs/njs_parser.c | 4 +- njs/njs_string.c

[njs] On-demand initialization of UTF-8 strings offset map.

details: http://hg.nginx.org/njs/rev/adf61ca4267b branches: changeset: 250:adf61ca4267b user: Valentin Bartenev date: Thu Nov 10 18:54:28 2016 +0300 description: On-demand initialization of UTF-8 strings offset map. diffstat: njs/njs_string.c | 70 ++--

[njs] Moved njs_string_offset_map_init() closer to its usage.

details: http://hg.nginx.org/njs/rev/0939df226d5c branches: changeset: 251:0939df226d5c user: Valentin Bartenev date: Thu Nov 10 19:09:13 2016 +0300 description: Moved njs_string_offset_map_init() closer to its usage. No functional changes. diffstat: njs/njs_string.c | 58 ++

Re: [PATCH] Core: slight optimization in ngx_chain_update_chains()

Hello! On Thu, Nov 10, 2016 at 10:30:34AM +0800, 胡聪 (hucc) wrote: > Hi, > > I found that sometimes the *out is NULL when i was reading the code > of ngx_event_pipe.c . So it is not necessary to traverse *busy when > *out == NULL in ngx_chain_update_chains(). > > # HG changeset patch > # User h

Re: [PATCH] Core: slight optimization in ngx_chain_update_chains()

Hello! On Fri, Nov 11, 2016 at 0:55 AM, Maxim Dounin wrote: >Looking into this again I tend to think that better solution would >be to test *out in additional if around all operations with *out, >like this (diff -w for clarity): > >@@ -186,6 +186,7 @@ ngx_chain_update_chains(ngx_pool_t *p, n >

Receiving sockets with SCM_RIGHTS

Hello! Recently I posted a patch for receiving sockets from another processes using SCM_RIGHTS messages. Since there is no observable activity on that, I want to discuss - is it a good idea to have such functionality in nginx core? May be there are some concepts, which I still don't know? Reg