Andrew Clayton
Andrew Clayton
Yeah, it can be done, but can be a bit fiddly... I'll remove my approval while I review the new stuff...
> Hi @ac000, > > I'm fine with both of the two styles when we do some features like this: a. split into two patches. > > ``` > 1....
> I’m afraid splitting into smaller patches doesn’t make review easier. It will be like this: > > 1. Introduce nxt_tstr_cond_t stuff. > > 2. Rework nxt_http_request_access_log() with nxt_tstr_cond_t. I...
> > 1. Introduce nxt_tstr_cond_t stuff. > > 2. Rework nxt_http_request_access_log() with nxt_tstr_cond_t. > > > I think it would. The patches would be smaller and more concise... > >...
Hi @andreivasiliu I've applied the typos and grammatical fixes [here](https://github.com/nginx/unit/commit/27c787f437ba94073d3349625c98c6d5c0a127ba) leaving the following for further contemplation. ```diff diff --git a/src/nxt_unit.h b/src/nxt_unit.h index a50046a5..1bec1094 100644 --- a/src/nxt_unit.h +++ b/src/nxt_unit.h @@ -119,19...
The way it works is that for each application we start a prototype process, this is then used to fork(2) off however many application processes are required. fork(2) (on Linux...
> 1. When a listener is removed, the controller process should not only > stop accepting new connections, but also stop keep-alive for > existing connections accepted by that listener....
OK, so here is something that seems to work. It will close the listen sockets and wait for the applications to finish. Some background... The router process is multi-threaded and...
> [@thresheek](https://github.com/thresheek) Would it be possible for SysEng to bump our Python language versions without a Unit release, or would that be a bit of a pain? 🙏 Not sure...
Hi @maaarghk Indeed, according to the [PSGI spec](https://metacpan.org/pod/PSGI) > The environment MUST include these keys (adopted from [PEP 333](http://www.python.org/dev/peps/pep-0333/), [Rack](http://rack.rubyforge.org/doc/files/SPEC.html) and [JSGI](http://jackjs.org/jsgi-spec.html)) except when they would normally be empty. and...