On Tue, Jun 28, 2016 at 1:52 PM, Jim Jagielski <[email protected]> wrote:
>
>> On Jun 27, 2016, at 4:25 PM, Yann Ylavic <[email protected]> wrote:
>>
>> On Mon, Jun 27, 2016 at 9:05 PM, Jim Jagielski <[email protected]> wrote:
>>>
>>>> On Jun 27, 2016, at 1:26 PM, [email protected] wrote:
>>>>
>>>> +    apr_bucket_brigade *tmp_bb;
>>>> } proxy_conn_rec;
>>>>
>>>
>>> I am missing the reason why this brigade needs to be
>>> a field in this struct. Is it simply to prevent us having
>>> to create it during each call of ap_proxy_check_backend()?
>>
>> Yes, mainly, we could create/destroy (yet another) temporary brigade
>> in ap_proxy_check_backend(), but there are several places already
>> where we do that (and where the proxy_conn_rec is available).
>> I thought we could later follow up on this change and optimize these
>> by using the new tmp_bb field (cleanup is faster than
>> create/destroy)...
>
> Oh. Gotcha. +1. We should make sure to document it as such.
> If there is data in the brigade between function calls, we
> will get some weird behavior :)

Good point, will do.

Reply via email to