On Fri, Mar 25, 2011 at 10:57 AM, Benoit Chesneau <[email protected]> wrote:
> oh right. So depending on the rule, vhost may be not working (simple
> are OK) . At this point I think it would be safer to remove any
> complex pattern matching and just match `host:/path` with only
> wildcard for the host for 1.1 . Other way would be to pass a full
> erlang term in the ini `vhosts = [{host, path}, ...]` but not sure it
> could be easily handled in the ini by the users.
Yep, sadly. I don't think any of the ets table types preserves
insertion order on retrievel. Just tried the bag type:
1> T = ets:new(ets_foo_bar_bar, [bag, private]).
16400
2> ets:insert(T, {b, 1}).
true
3> ets:insert(T, {a, 2}).
true
4> ets:insert(T, {x, 3}).
true
5> ets:insert(T, {f, 4}).
true
6> ets:tab2list(T).
[{x,3},{f,4},{b,1},{a,2}]
7> ets:first(T).
a
I think our .ini parser also might not insert the .ini key/values in
the same order as they're listed in the files, but i haven't
confirmed.
>
> Maybe first solution is better, and we could go for a webmachine
> version of http in 1.2 (or 2?) I can work on that.
Perfectly fine for me.
>
> - benoit
>
--
Filipe David Manana,
[email protected], [email protected]
"Reasonable men adapt themselves to the world.
Unreasonable men adapt the world to themselves.
That's why all progress depends on unreasonable men."
