doc: limit_except

[Gregg Reynolds]

On 3/16/07, Igor Sysoev <is at rambler-co.ru> wrote:

>
> Although initially "limit_except" was used for authorization only,
> now it allows two non-authorization directive - "proxy_pass" and "perl"
> and can be use as method switches as well as "location"s are used as
> URI swicther. If I will make support of several "limit_except" blocks
> the directive should be even renamed to "methods_except":
>
>          location / {
>
>               methods    POST {
>                    proxy_pass ...
>               }
>
>               methods    LOCK  UNLOCK {
>                    perl  ...
>               }
>
>               methods_except    GET {
>                    ...
>               }
>
>               # GET/HEAD
>               ...
>          }
>
I like it.  But why not use logical ops?  e.g.

>          location / {
>
>               methods ==  POST {
>                    proxy_pass ...
>               }
>
>               methods ==  LOCK  UNLOCK {
>                    perl  ...
>               }
>
>               methods !=  GET {
>                    ...
>               }
>
>               # GET/HEAD
>               ...
>          }

One advantage is the logical operator clearly only applies to the
argument, not to the block, and it avoids the messy natural language
semantics of "except".

Thanks for the explanation - it's very useful to know the motivating
problem behind the design.

-gregg