> Also, I was never 100% decided about one aspect of the bucket copy API, which > is the > return value... should the ap_bucket* be the actual return value, or should > it be an > apr_status_t? I ended up going with the apr_status_t return value to > accomodate > APR_ENOTIMPL. Do people agree with that decision?
It needs to be apr_status_t to accomodate ENOTIMPL and ENOMEM. > On the other hand, it seems that ap_brigade_copy() ought to return an > ap_bucket_brigade* to match ap_brigade_split() and ap_brigade_split_offset(). > Yes? Since brigade_split_offset can't be ENOTIMPL, we can return a brigade. If there is no memory, just return a NULL brigade. Ryan _______________________________________________________________________________ Ryan Bloom [EMAIL PROTECTED] 406 29th St. San Francisco, CA 94131 -------------------------------------------------------------------------------
