[EMAIL PROTECTED] wrote:
>
> On Tue, 3 Jul 2001, jean-frederic clere wrote:
>
> > Yes, but jk_open_socket would have to be modified:
> > 1 - to pass a apr_pool to allocate the apr_connect.
> > 2 - to return a status instead a socket.
>
> Sure, you can pass a jk_pool ( and make jk_log a field of jk_pool, so
> everything will work as before ), and create a structure that mimics
> apr_socket_t. ( that if you want to use the #define jk_open_socket apr_...
> and #define jk_pool apr_pool ).
>
> Then convert the 2-3 places where the function is called - the code will
> still work almost the same way as before, and we'll be a step closer to
> switch to APR when we're ready ( and everything will still work ).
>
> The only issue - if you decide to start this, it would be good to either
> wait 1-2 weeks, we need a stable mod_jk for 3.3 beta ( we can tag the
> workspace when we release the beta and keep it stable for the release ).
No problem - I have also to help with scandoc as I proposed it some days ago -
>
> > My idea is to use this mechanism eveywhere.
>
> But slowly :-)
>
> +1
>
> Costin
